TEXT FLY WITHIN
THE BOOK ONLY
Lu<OU 164120 >m
OSMANIA UNIVERSITY LIBRARY
lor
This book should be returned on or before the date
" * *%
i.st marked below. .
McGRAW-HILL PUBLICATIONS IN THE
AGRICULTURAL SCIENCES
LEON J. COLE, CONSULTING EDITOR
METHODS
of
PLANT BREEDING
The Quality of the materials used in the
manufacture of this book is governed by
continued post-war shortages.
SELECTED TITLES FROM
McGRAW-HILL PUBLICATIONS IN THE
AGRICULTURAL SCIENCES
LEON J. COLE, Consulting Editor
Adriance and Brison - PKOPAGATJON OF HORTICULTURAL PLANTS
Brown COTTON
Cruess COMMERCIAL FRUIT AND VEGETABLE PRODUCTS
EckleSj Combs, and Macy MILK AND MILK PRODUCTS
Fawcett CITRUS DISEASES %
Fernald and Shepard - APPLIED ENTOMOLOGY
Gardner, Bradford, and Ilojkcr FRUIT PRODUCTION
Gustafson - CONSERVATION OF THE SOIL
Gustafson SOILS AND SOIL MANAGEMENT
Hayes and Garber - BREEDING CROP PLANTS
Hayes and Immer - METHODS OF PLANT BREEDING
Heald - MANUAL OF PLANT DISEASES
Ileald INTRODUCTION TO PLANT PATHOLOGY
Hutcheson, Wolfe, and Kipps FIELD CROPS
Jenny - FACTORS OF SOIL FORMATION
Jull - POULTRY HUSBANDRY
Laurie and Ries - FLORICULTURE
Leach INSECT TRANSMISSION OF PLANT DISEASES
Little and Plastridge BOVINE MASTITIS
Maynard ANIMAL NUTRITION
Metcalf and Flint DESTRUCTIVE AND USEFUL INSECTS
Paterson STATISTICAL TECHNIQUE IN AGRICULTURAL RESEARCH
Peters - LIVESTOCK PRODUCTION
Rather FIELD CROPS
Rice - BREEDING AND IMPROVEMENT OF FARM ANIMALS
Roadhouse and Henderson THE MARKET-MILK INDUSTRY
Robbins, Crafts, and Raynor WEED CONTROL
Schillctter and Richey TEXTBOOK OF GENERAL HORTICULTURE
Thompson - VEGETABLE CROPS
Waite POULTRY SCIENCE AND PRACTICE
There are also the related series of McGraw-Hill Publications in
the Botanical Sciences, of which Edmund W. Sinnott is Consulting
Editor, and in the Zoological Sciences, of which A. Franklin Shull
is Consulting Editor. Titles in the Agricultural Sciences were
published in these series in the period 1917 to 1937.
METHODS
of
PLANT BREEDING
BY
HERBERT KENDALL HAYES
Professor and Chief of Division of Agronomy and
Plant Genetics, College of Agriculture,
University of Minnesota
AND
FORREST RHINEHART IMMER
Professor of Agronomy and Plant Genetics,
College of Agriculture, University of Minnesota
FIRST EDITION
FIFTH IMPRESSION
McGRAW-HILL BOOK COMPANY, INC.
NEW YORK AND LONDON
METHODS OF PLANT BREEDING
COPYRIGHT, 1942, BY THE
McGRAW-HiLL BOOK COMPANY, INC.
PB1NTED IN THE UNITED STATES OF AMEK1CA
All rights reserved. This book, or
parts thereof, may not be reproduced
in any form without permission of
the publishers.
COMPOSITION BY THE MAPLE PRESS COMPANY, YORK, PA,
PRINTED AND BOUND BY COMAC PRESS, INC., BROOKLYN, N. Y,
PREFACE
Plant breeding is an applied science that is carried out effici-
ently only through the application of other basic plant sciences.
The rapid increase in knowledge of genetics since the rediscovery
of Mendel's laws of heredity in 1900 and the application of these
laws to plant breeding were essential steps in the development of
plant breeding as a science. The contributions of cytogenetics
in recent years have furnished, in many cases, a clear picture of
genetic relationships based upon differences and similarities of
chromosome morphology, structure, arid function. Many eco-
nomic plants are polyploids, and a knowledge of chromosome
numbers, pairing behavior in crosses, and gene differences among
related species and varieties is essential in building new varieties
of plants with the characters desired by the grower and con-
sumer. Physical and chemical methods of inducing changes in
chromosome number and structure and of inducing gene changes
are being developed. Satisfactory technics for inducing poly-
ploidy in species and hybrids are available for certain types of
plant breeding problems.
In order to evaluate a variety, it is necessary to compare it
with varieties of known performance. The comparisons made
by the plant breeder are extensive, and frequently only a few
replications can be grown. The development of adequate
statistical methods has aided greatly in making reliable com-
parative trials. Experimental methods of making reliable
comparisons are one of the tools of the plant breeder.
Methods have been devised in many cases for differentiating
quality, for a determination of the relative value of different
characters, including chemical properties, that make it possible
, under conditions of controljjadf* pollination to select for the
characters desired. In problems of breeding for disease resis-
tance, a knowledge of the genetics of the pathogen is as essential
as that of the crop plant itself. With each individual plant,
information regarding available varieties, their characters, and
vi PREFACE
their wild relatives furnishes a basis for the combination of genes
desired by the breeder. For diseases caused by pathogens it is
equally important to know the probable mode of origin of new
strains of the organism, and the number, distribution, and genetic
nature of the strains present in the region where the crop plant
is to be grown.
The subject matter presented in " Methods of Plant Breed-
ing" has been used in both undergraduate and graduate courses
at the University of Minnesota. The undergraduate course is
taught only to junior and senior students. The graduate courses
are given for the purpose of teaching standardized methods of
breeding for particular categories of breeding problems and to
present the current viewpoint when the most desirable method
of breeding is not so well known. This is with the belief that
each of the various methods of hybridization, including the
pedigree method of selecting during the segregating generations,
the bulk method with self-pollinated plants, the backcross
method and convergent improvement, has certain advantages and
disadvantages that make it desirable under some conditions
and less desirable for other breeding problems.
A great deal of information is available regarding the genetics
of many crop plants, and added information is being obtained
very rapidly. It seems unwise to attempt a complete review
of the present status of the genetics of many crop plants, since
the information available is very extensive and such a review
would be out of date almost as soon as it was published. Concise
reviews of the mode of inheritance of important characters of the
small grains, flax, and corn have been included to illustrate the
value to the breeder of a knowiedgc of inheritance as an aid in
planning the breeding program. These should be supplemented
by similar reviews of inheritance for those crop plants that are
of greatest value for each class of students who use the book.
The present status of corn breeding, a rather typical cross-
pollinated plant, has been reviewed in considerable detail, since
many of the studies made with corn and the results obtained are
basic to an understanding of principles of breeding other cross-
pollinated plants.
Methods of field-plot technic, experimental design, and statis-
tical analysis with particular reference to plant-breeding problems
have been discussed, including some of the newer methods. The
PREFACE vii
necessary statistical tables have been included with the permis-
sion of the original publishers.
The authors are indebted to Professor R. A. Fisher and to
Messrs. Oliver and Boyd, of Edinburgh, for permission to reprint
completely or in abridged form Appendix Tables I ? III, and IV
from their book " Statistical Methods for Research Workers,"
7th Ed. (1-938) and to Professor George W. Siiedecor and his
publishers, Iowa State College Press, for permission to reprint
Appendix Table II from their book "Statistical Methods/' 3d
Ed. (1940). Professor Snedecor has given permission to reprint
Appendix Table V, and Dr. C. I. Bliss has given permission to
reprint Appendix Table VI.
Various co workers have read particular chapters and have
made helpful suggestions. Particular thanks are due to Dr.
0. R. Burnham for suggestions regarding the chapters on Gen-
etics and on Inheritance in Maize; to Dr. E. R. Ausemus for
suggestions regarding the chapter on Inheritance in Wheat;
to Dr. F. A. Krantz for helpful suggestions regarding potato
improvement; to A. G. Tolaas for information regarding potato-
seed certification; and to Dr. C. H. Goulden for reviewing the
chapters on Field-plot Technic and on Statistical Methods. Dr.
H. M. Tysdal kindly furnished unpublished information regarding
the effects of self-pollination in alfalfa. In problems relating to
disease resistance, suggestions by Dr. J. J. Christensen and M. B.
Moore have been specially helpful. " Breeding Crop Plants/' by
Hayes and Garber, has been used freely. The writers, however,
accept full responsibility for the viewpoints presented.
H. K. HAYES,
F. R. IMMER.
UNIVERSITY or MINNESOTA,
February, 1942.
CONTENTS
PAGE
PREFACE: v
CHAPTER I
THE ROLE OF PLANT BREEDING , 1
The value of plant breeding -Genetic principles are the basis of
scientific breeding Breeding spring wheat resistant to stem rust >
Corn breeding Potato improvement.
CHAPTER II
THE GENETIC AND CYTOGENETIC BASIS OF PLANT BREEDING 11
Chromosome number (haploid) in common crop plants Polyploids
in relation to plant breeding Some applications of genetics to
plant breeding Colchicine as a polyploidizing agent.
CHAPTER III
MODE OF REPRODUCTION IN RELATION TO BREEDING METHODS. ... 39
The asexual group The sexual group Self-pollination leads to
homozygosis The effects of self-pollination in the often cross-
pollinated group. Effects of self-fertilization in cross-pollinated
plants Heterosis and its explanation A classification of methods
of breeding sexually propagated plants.
CHAPTER IV
TECHNICS IN SELFING AND CROSSING 60
Corn Wheat, oats, and barley Rye Flax Cotton Sorghum
Rice Potato Pumpkin and squash Onion Red Clover Al-
falfa and sweet clover Grasses.
CHAPTER V
THE PURE-LINE METHOD OF BREEDING NATURALLY SELF-POLLINATED
PLANTS 74
Early studies The pure-line theory The pure-line theory in its
application Methods of improving self-fertilized plants by indi-
vidual-plant selection Utilization of introductions- Pedigree
selection within adapted varieties Cooperative tests and dis-
tribution of promising lines Illustrations of valuable varieties of
self-pollinated plants produced by application of the pure-line
theory.
CHAPTER VI
HYBRIDIZATION AS A METHOD OF IMPROVING SELF^POLLINATKD PLANTS 86
Some studies before 1900 Development of methods since 1900
Breeding improved varieties of barley Breeding by hybridization
ix
X CONTENTS
PAGE
Object of crowing Selection of parental material Technic of
crossing Handling the hybrid material Methods of breeding
Pedigree method Bulk method Backcross method Multiple
crosses Combining ability.
CHAPTER VII
THE BACKCROSS METHOD OF PLANT BREEDING 101
Genetic expectations from backcrossmg Cantaloupes resistant to
powdery mildew, Erysiphe cichoracearum Breeding bunt-resistant
wheats Breeding rust-resistant snapdragons Studies at the
Minnesota station Disease resistance in wheat Illustrations with
corn.
CHAPTER VIII
BREEDING FOR DISEASE AND INSECT RESISTANCE 113
The importance of disease resistance Methods of breeding for
disease and insect resistance The search for resistant material
Artificial production of ephiphytotics Black stem rust of wheat,
oats, and barley (Puctinia graminis) Leaf rust of wheat (P.
triticina) and crown rust of oats (P. coronata) Bunt of wheat
(Tilletia tritici) Loose smut of oats (Ustilago avenae); covered
smut of oats (U. levis) Covered smut of barley (U. hordei) and
intermediate smut of barley (U. medians) Flax rust (Melamp-
sora lini) Flax wilt (Fusarium lini) Fusarial head blight (scab)
of wheat and barley Corn smut (17. zeae) Loose smut of sor-
ghum (Sphacelotheca cruenta) and covered smut of sorghum ($.
sorghi) Head smut of sorghum and corn (S. reilianum) Hessian-
fly injury of wheat Methods of breeding Study of fundamental
problems,
CHAPTER IX
INHERITANCE IN WHEAT 129
Glume shape Awnedness Chaff characters Seed characters
Spike density Spring and winter habit Stem-rust reaction
Bunt resistance Other problems of disease resistance Quanti-
tative characters.
CHAPTER X
INHERITANCE IN OATS 141
Species groups Inheritance of characters in crosses between 42
chromosome species Differences in awn development Color of
grain Hulled vs. hull-less Spreading vs. side panicle Pubes-
cence Disease reactions (Stem rust Crown rust Correlated
inheritance of reaction to ' three diseases Smuts Quantitative
characters.
CHAPTER XI
INHERITANCE IN BARLEY " . . . . 152
Classification and genetics of barley species Chromosome number
in genus Hordeum Linkage groups Internode length in the
CONTENTS xi
PAGE
rachis of the spike Reaction to Helminthosporium sativum
Reaction to stem rust Resistance to mildew Interaction of
factors affecting quantitative characters.
CHAPTER XII
INHERITANCE IN FLAX 165
Factors for flower and seed color in common flax Dehiscence of
the bolls Smooth vs. ciliate septa Weight of seed and oil content
Inheritance of quality of oil Disease resistance Wilt resis-
tance Resistance to rust.
CHAPTER XIII
METHODS OF SELECTION FOR SPECIAL CHARACTERS ......... 175
Quality tests in wheat Wheat-meal fermentation time test
Cold-resistance tests with wheat Shattering in wheat Dor*
mancy in relation to breeding Lodging in small grains and corn
Drought studies with corn Inducing biennial sweet clover to
flower the first year Determination of coumarin content in sweet
clover Method of determining hydrocyanic acid content of
single plants of sudan grass.
\ CHAPTER XIV
DEVELOPMENT OF METHODS OF CORN BREEDING 187
Selection without controlled pollination Early studies of self-
and cross-fertilization with com Controlled pollination methods
Breeding improved inbred lines The pedigree method of selection
in the segregating generations after crossing inbreds Genetic
diversity The backcross Convergent improvement.
CHAPTER XV
INHERITANCE IN MAIZE 215
Origin and classification The pod corns The flint corns The
popcorns The dent corns The flour corns The sweet corns
The waxy corns Endosperm characters Chlorophyll variations
Plant color Glossy seedlings Linkage studies with maize
Inheritance of quantitative characters Linkage of factors for
row number with genes at known loci Inheritance of smut reac-
tion Inheritance of combining ability Inheritance of other
important characters.
CHAPTER XVI
CONTROLLED POLLINATION METHODS OF BREEDING CROSS-POLLINATED
PLANTS 242
Effects of self-fertilization Inheritance of self-incompatibility *
Methods of breeding Outline for improvement of cross^poUinated
plants by controlled pollination methods.
xii CONTENTS
PAGE
y CHAPTER XVII
SEED PRODUCTION 263
Selecting the variety First increase of seed of a new variety
Seed certification and registration The Canadian Seed Growers'
Association The International Crop Improvement Association
Description of seed classes The Minnesota plan for certain crops
Seed (tubers) certification for potatoes.
CHAPTER XVIII
SOME COMMONLY USED MEASURES OF TYPE AND VARIABILITY .... 280
Definition of statistical constants Calculation of mean Standard
error, variance, and coefficient of variability Correlation coef-
ficient Comparison of differences by the t test.
CHAPTER XIX
FIELD-PLOT TECHNIC 289
Crop rotation for experimental fields Soil heterogeneity Com-
petition Size and shape of plots Replication Methods of
making yield trials used in Minnesota.
CHAPTER XX
RANDOMIZED BLOCKS, LATIN SQUARES, AND x 2 TESTS 307
Tests in randomized blocks Latin squares Estimating the yield
of a missing plot Split-plot experiments Chi square (x 2 ) tests.
CHAPTER XXI
CORRELATION AND REGRESSION IN RELATION TO PLANT BREEDING . . . 325
Simple correlation Linear regression Means and differences of
correlation coefficients Partial correlation Multiple correlation.
CHAPTER XXII
MULTIPLE EXPERIMENTS, METHODS OF TESTING A LARGE NUMBER OF
VARIETIES, AND THE ANALYSIS OF DATA EXPRESSED AS PERCENTAGES 339
Multiple experiments in randomized blocks Comparison of varie-
ties in different experiments where the same check varieties are
grown Simple lattice experiments Triple lattice experiments
Analysis of data expressed as percentages.
LITERATURE CITATIONS 3H1
if
GLOSSARY 401
APPENDIX 411
INDEX 421
METHODS OF PLANT BREEDING
CHAPTER I
THE ROLE OF PLANT BREEDING
There is a growing appreciation of the value of plant breeding
as<a means of obtaining new or improved plant forms adapted to a
wide variety of uses. Although most important food plants had
been brought under cultivation before the dawn of recorded
history, there remains at present almost an unlimited opportunity
to improve the varieties of plants available for various agricul-
tural uses and in some cases greatly to modify their characters.
The primary purpose is to obtain or produce varieties or hybrids
that are efficient in their use of plant nutrients, that give the
greatest return of high-quality products per acre or unit area in
relation to cost and ease of production, and that are adapted to
the needs of the grower and consumer. It is of great importance
also to obtain varieties that are able to withstand extreme
conditions of cold or drought or that have resistance to patho-
genic diseases or insect pests. Such qualities help materially to
stabilize yields by controlling extreme fluctuations.
Although the art of plant breeding, i.e., the ability to discern
fundamental differences of importance in the plant material
available and to select and increase the more desirable types, is a
great asset to the breeder, there is general appreciation that plant
breeding, efficiently carried out, is to a large extent dependent on
fundamental training in the biological sciences.
Some of the more important phases of this training may be
summarized as follows:
1. A knowledge of genetic and cytogenetic principles.
2. A knowledge of the characteristics of the crop to be
improved, including its wild relatives.
3. Information regarding the needs of the grower.
4. A knowledge of special technics adapted to the solution of
particular problems.
1
2 METHODS OF PLANT BREEDING
5. A knowledge of the principles of field-plot technic.
6. A knowledge of the principles involved in the design of
experiments and the statistical reduction of data.
The purpose of this book is to summarize the methods of breed-
ing that have been developed for particular categories of crop
plants, to explain the reason why particular methods are chosen
for certain types of crop-improvement problems, and to give
methods of field-plot technic and of statistical analysis that are
adapted for particular uses. Although individual crop problems
will be used for illustration, there will be no attempt to summarize
the work that has been done or that needs to be done with
individual crop plants except as these facts may aid in under-
standing types of problems. Emphasis will be placed on methods
of breeding and principles underlying their use.
THE VALUE OF PLANT BREEDING
E. F. Gaines (1934), of the Washington Agricultural Experi-
ment Station, has made the statement that the practical results of
genetic research on disease resistance in plants has helped to
popularize genetics with the general public. The illustration of
breeding stem-rust-resistant wheat given later in the chapter
emphasizes the value of cooperation in research and the need of
intensive study in order to solve underlying principles and thus
make possible the solution of complex problems and the develop-
ment of the desired varieties.
In discussing the importance of a knowledge of genes and their
control, the following statements have been made by Muller:
Organisms are found to be far more plastic in their hereditary basis
than has been believed, and we may confidently look forward to a future
in which if synthetic chemistry shall not have displaced agriculture
the surface of the earth will be overlaid with luxuriant crops, at once
easy to raise and to gather, resistant to natural enemies and climate,
and readily useful in all their parts.
This work is a far vaster one th^n the layman ordinarily realizes, for
there are many thousands of wild species of plants whose varied potenti-
alities must be tested, and many species both wild and cultivated already
contain hundreds of varieties and thousands of individual differences.
By means of laborious crossing methods, these diverse types may be
combined and recombined within wider limits, and so a virtually endless
succession of specialized hybrid forms may be produced, differentiated
THE ROLE OF PLANT BREEDING 3
into local geographical races each having characteristics especially suited
to its peculiar conditions of cultivation and to the needs of the district.
When to the potentialities of hybridization are added those that will
appear as new hereditary types arising by mutation, the path of change
and adaptation is seen to be indeed limitless.
GENETIC PRINCIPLES ARE THE BASIS OF SCIENTIFIC BREEDING
Many years ago Raymond Pearl emphasized the fact that with
self-pollinated crop plants the plant breeder was using Mendel's
laws as a direct working gilide. Illustrations by the score could
be given to show how particular types of genetic knowledge have
been and are being used as a basis for a planned crop-improve-
ment program. A few illustrations will be given to show the
extent to which a knowledge of the genetics of a particular crop is-
essential in the development of a logical breeding program.
Breeding Spring Wheat Resistant to Stem Rust. One of the
principal cooperative projects at Minnesota since 1915, in which
agronomists, plant geneticists, cereal chemists, and plant path-
ologists have all played their parts, has been the development of
rust-resistant varieties of spring wheat of desirable agronomic
type and of satisfactory milling and baking quality. This
research program has been carried on through cooperation
between workers in the Minnesota Experiment Station and the
U. S. Department of Agriculture.
In these studies, artificial epidemics of stem rust have been
developed both under field conditions and in the greenhouse.
The rust nursery in the field has consisted of several thousand
rows yearly. During the early period of this study, resistant
vulgare wheats were unknown. The present nursery has sudi a
preponderance of strains of vulgare wheats highly resistant to
stem rust that it is necessary to plant a considerable amount of
susceptible host material throughout the nursery in order that
rust may develop sufficiently so that a satisfactory spread of the
disease may be made possible. The development of rust-
resistant strains has been accomplished by obtaining resistance
from the Emmer group and by combining this resistance with the
desirable agronomic characters of vulgare wheats through a series
of crosses and selections.
At the present time much remains to be known about various
phases of stem-rust resistance in wheat, but many problems have
4 METHODS OF PLANT BREEDING
been solved. Some of the steps leading to our present position
may be mentioned.
1. The mode of inheritance of particular types of reaction to
stem rust has been determined in both the greenhouse and field.
The most important practical result of these studies is the con-
clusion that resistance to all races of stem rust of wheat in the
stage from heading to maturity may be dependent upon only a
single or a few genetic factors.
2. The pathogene causing the disease Puccinia graminis tritici
Eriks. & Henn. is composed of numerous forms, called physiologic
races, that can be differentiated by their manner of reaction
with a series of wheat varieties and species known as differential
hosts, this separation being made primarily on the basis of
seedling reaction. A wheat variety resistant to a particular race
of rust in the seedling stage is resistant to the same race in all
stages of plant growth under field conditions. Physiological
resistance in the seedling stage is of such a nature that a wheat
may be immune from one race of rust and susceptible to another.
As an illustration, Kanred winter wheat and some hybrid deriva-
tives having Kanred as an ancestor are immune from certain races
and highly susceptible to others. This knowledge explains the
reason why Kanred winter wheat and derivatives may be highly
resistant in one season and highly susceptible in another.
3. A knowledge of the causes of resistance has been of major
importance. Thus, the resistance of Kanred is physiological and
acts only against particular races of rust. A second type of
resistance under field conditions to many races of rust as the
plants approach maturity, called mature-plant resistance, appears
to be simply inherited. The exact cause of this type of resistance
is unknown. Some have suggested that morphological and func-
tional causes may be responsible. Others have given evidence
indicating that this does not seem to be the explanation. Mature-
plant resistance is inherited, in some cases, in a simple Mendelian
manner, especially where the varieties Hope and H44 are used as
the resistant parents. f ;
4. It has been learned also that extreme conditions of environ-
ment may cause an apparent breaking down of resistance to a
particular disease. For example, a plant genotypically resistant
to stem rust, if infected with loose smut, may be completely
susceptible to rust. This conclusion seems essential in a logical
THE ROLE OF PLANT BREEDING 5
viewpoint of disease resistance in plants. No one expects that a
potentially high-yielding variety will give high yields under
unfavorable conditions. Extreme conditions of environment
may strongly modify reaction to disease by modifying the charac-
ter that, under normal conditions, is responsible for the resistance
to that particular disease.
FIG. 1. Thatcher wheat was first released in Minnesota in the spring of
1934. There were approximately 12,000,000 acres grown in Canada in 1940 arid
5,500,000 acres in the United States. The estimate has been made that Thatcher
has given an annual increase in farm income to the Minnesota farmer of 2 million
dollars.
Thatcher wheat (Hayes et al. 1936), first introduced in the
spring of 1934 in Minnesota, is now the most widely grown stem-
rust-resistant wheat, being the major spring wheat grown in 1939
in the United States in the eastern and central sections of the
spring-wheat area. Thatcher is grown extensively also in
Canadian provinces where stem rust is most severe. It withstood
the stem-rust epidemics of 1935, 1937, and 1938 when susceptible
varieties of spring wheats were very severely injured. Thatcher
excels in yielding ability, strength of straw, and milling and
baking quality but is somewhat less satisfactory in weight per
6 METHODS OF PLANT BREEDING
bushel than some other varieties, partly because of its small
size of seed and its susceptibility to scab and leaf rust. The
latter disease was epidemic in 1938 in the spring-wheat area.
Thatcher is the product of a double cross between (lumillo
durum X Marquis) X (Marquis X Kanred). From the first
cross of lumillo X Marquis and selection during F% to F^ 4 related
.strains were obtained with 21 pairs of chromosomes, vulgare type
of plant, and resistance to stem rust. It is of interest that no
plants were obtained in F% that had this combination of characters
but that from over one thousand F 3 lines there was one line that
contained several plants that were of vulgare type that were
resistant to stem rust. From the cross of Marquis X Kanred, a
considerable number of spring wheats were selected that were
homozygous for the immunity of Kanred to several rust races.
Thatcher was selected from a cross between the more promising
strains of these two single crosses. The Thatcher variety com-
bines field resistance to many physiologic races of stem rust
with seedling resistance to the races to which Kanred is immune.
It was the first successful attempt to transfer rust resistance from
the Emmer group with a haploid chromosome number of 14 to the
vulgare group (n = 21).
McFadden (1930) was the first to develop vulgare wheats
with near immunity to stem rust from crosses of Marquis with
Yaroslav emmer. He produced two wheats, Hope and H44, with
42 chromosomes that in the mature-plant stage are highly resist-
ant to stem rust. Neither of these wheats is entirely satisfactory
in other characters. The resistance of Hope and H44 to all rust
races in the field under normal conditions in North America is
dependent upon one or two major genetic factors for resistance.
Most of the more promising new spring wheats have Hope or H44
somewhere in their parentage.
Corn Breeding, The viewpoint . has been expressed by various
writers that the production of adapted corn hybrids for different
regions of the corn belt will have the most far-reaching effect of
any phase of work in crop improvement of the present generation.
The hybrid Burr-Learning was first distributed in Connecticut in
1922, but the acreage grown of this hybrid has been very small.
The first distribution of hybrids in the corn belt occurred from
1932 to 1934, and in 1938 and 1939 from 15 to 25 million acres
were planted to hybrid corn, leading to an increased production
THE ROLE OF PLANT BREEDING 7
of from 100 to 150 million bu. of corn over what would have been
obtained if hybrid corn had not been available. Many agrono-
mists believe that there will continue to be a rapid increase in the
use of hybrid corn in the years to come until the greater part of
the acreage of corn in the United States is planted to hybrid
varieties.
The widespread interest in hybrid corn is due primarily to the
superiority of hybrids over normal varieties in a number of char-
acters. Although higher yields per acre are important, other
improvements are of equal and perhaps greater value. Ability to
withstand lodging and resistance to smut and to ear and stalk rots
are of major importance. The development of drought-resistant
and frost-resistant hybrids has been studied also, although much
remains to be accomplished in these fields.
In the development of hybrid corn, Mendelian principles have
been used directly. A standardized technic of breeding has been
developed based on the direct application of principles of genetics.
Intensive studies of inbreeding and crossbreeding corn were
started by E. M. East at the Connecticut Agricultural Experi-
ment Station and G. H. Shull at Cold Spring Harbor in 1905.
Many investigators have taken part in studies of inheritance in
maize. The fundamental principles elucidated have led to a
sound basis for scientific improvement in corn, a field in which a
considerable number of investigators devote all or part of their
research efforts. Some of the more important principles leading
to the present methods will be mentioned, although corn breeding
will be outlined in detail in a later chapter.
1. Continued self-pollination in corn leads to the production of
relatively homozygous types that are in general less vigorous than
normal corn. Crossing inbreds restores vigor. Some FI crosses
are more vigorous than normal corn; others, less so.
2. Crosses between inbreds are difficult to use in commercial
seed production, since the yield of seed per acre is low. This
difficulty has been overcome by using for commercial production
crosses between single crosses.
3. Hybrid vigor in corn and in other crop plants has been
placed on a definite Mendelian basis. It is a result of partially
dominant growth factors. Many genes are involved in growth
vigor, and consequently linkage makes it difficult to combine all
important genes in a single inbred line.
8 METHODS OF PLANT BREEDING
4. Some inbred lines have much better combining ability than
others when tested in comparable crosses. By crossing a group of
inbreds to be used in a definite breeding program with a variety
and by testing the inbred-variety crosses in yield trials, the
better combining inbreds can be isolated and the less desirable
discarded.
5. The combining value of inbred lines in a double cross can be
predicted from yield trials of the appropriate single crosses.
From each of four inbred lines, six single crosses and three double
crosses can be made. Yields of any particular double cross can
be predicted from the average yield of the four single crosses not
used in making the double cross. These results may be under-
stood on the basis that a double cross produced from advanced
generations of two single crosses behaves approximately the same
as the double cross between the two single crosses.
6. The ease of commercial production of double-crossed seed is
dependent to a considerable extent upon the vigor of the inbred
lines as well as the yielding ability of the single crosses used in
the double cross. Improved inbred lines of corn can be bred by
the same breeding methods as used in the production of improved
varieties of self-pollinated plants, although it is necessary to
control pollination by appropriate selfirig and crossing in carrying
out the program.
7. The principles of corn breeding that make possible the
utilization of hybrid vigor are dependent upon an understanding
of genetic principles and their application to corn breeding. This
knowledge has made possible to a considerable extent the stand-
ardization of corn-breeding technics.
Potato Improvement. Since commercial varieties of potatoes
are highly heterozygous, plants grown from seed will vary greatly.
The selection and clonal increase of plants developed from seed
was naturally the first breeding method used and led to the
production of the old standard varieties. During the early
period of the present century, selection within clones was used as
a method of potato improvement. Although of little value in
breeding, clonal selection was an aid in isolating plants free from
virus diseases. This led to the use of the tuber-unit method in the
diagnosis for and eliminating of plants affected by virus diseases.
Most varieties of the potato are nonself-fruitful, but strains
have been isolated that are highly self -fruitful, Clones of the
THE ROLE OF PLANT BREEDING 9
latter type produce an appreciable amount of stainable pollen.
The partial pollen sterility found in the self-fruitful clones appears
to be due to abortion of the pollen grains after regular meiotic
division. In the nonself-fruitful clones very little stainable
pollen is produced because of irregular meiotic division. In
crosses between self- and nonself-fruitful clones, the progeny are
usually highly nonself-fruitful.
Improvement of self-fruitful clones may be accomplished
through selfing and selection withiii clones and crosses between
clones, with the use of the breeding methods applicable to self-
fertile crops. By the use of such methods clones resistant to late
blight, scab, and specific virus diseases have been produced. The
breeding value of selections in self-fruitful clones may be deter-
mined by tests of clonal progenies or, preferably, by tests of the
selfed progeny. Thus, two clones may appear to produce the
same amount of disease but be found to differ in genotype when
selfed progenies of these clones are compared. The genotype of
clones that are pollen-sterile may be determined from crosses with
pollen-fertile clones of known genotypes by the use of the pollen-
sterile clones as females.
The use of inbreeding methods, supplemented by planned
crosses, has resulted in a rapid increase in knowledge of the
genetics of the potato and the isolation of superior germ plasm in
this crop. In the utilization of this superior germ plasm, some
modifications in the ordinary breeding, however, must be made in
the methods applicable to self-fertile crops.
Although self -fertility is of value in the isolation and synthesis
of strains that are resistant to disease or insect attack and have
desirable agronomic characters, self-fruitfulness in itself leads to a
loss in yield of tubers. Plants that produce flowers and fruits
yield less than plants that do not, the reduction in yield being
proportional to the number of flowers or fruits produced. Conse-
quently, the character of self-fruitfulness, after being utilized
during the production of superior strains, must be eliminated in
the breeding of commercial varieties. This may be accom-
plished through crosses between nonself-fruitful commercial
varieties possessing high yielding capacity and certain plant and
tuber types with self -fruitful clones that possess the characters to
be added, with the use of the self-fruitful clones as pollen parents.
The FI progeny of such crosses are highly nonself-fruitful.
10 METHODS OF PLANT BREEDING
Selection of nonfruiting types in these progenies may be expected
to result in improved varieties into which have been synthesized
the characters desired.
The use of such methods of breeding has resulted in the develop-
ment and release to the growers of more than a dozen new varie-
ties during the past six years. All these have as one parent at
least a superior pollen parent developed at the Minnesota Agricul-
tural Experiment Station or by the U, S. Department of Agricul-
ture. Some of these varieties are resistant to late blight; others
are resistant to specific viruses, common scab, or insect attack.
Warba is a newly developed early maturing variety that is
resistant to mosaic and possesses high-yielding ability. The
Sebago variety withstood the severe late blight epidemic of 1938
remarkably well. Katahdin, with its viable pollen, provides the
plant breeder with a high-yielding commercial variety that can be
used as a pollen parent in further breeding. The synthesis of
commercial varieties that have resistance to several diseases, as
well as many desirable agronomic characters, is preceeding
rapidly.
CHAPTER II
THE GENETIC AND CYTOGENETIC BASIS OF
PLANT BREEDING
A knowledge of the chromosome basis of heredity is essential to
the breeder. The characters of a plant are the end result of the
interaction of genes, carried in the chromosomes, under particular
environmental conditions. What is inherited is the manner of
reaction and not the character itself.
Diploid organisms result from the union of male with female
reproductive cells, the chromosome number in the zygote nor-
mally being twice that of the gamete. With self-pollinated
organisms, homozygosis is obtained automatically, and perma-
nence of characters under uniform conditions of environment is
obtained as a result of equational division of each chromosome
and gene during somatic mitosis, the diploid conditions of the
chromosomes in the body and the pairing during reduction divi-
sion of like chromosomes, two by two, leading to the production
of a single genetic type of gamete.
The linear arrangement of genes in the chromosome has been
generally accepted, and the division of the gene in mitosis and its
segregation in meiosis have furnished the mechanism for the
transfer of the unit of inheritance, the gene, from cell to cell. A
knowledge of the number and nature of the chromosomes in each
crop plant and their behavior in cell division is fundamental to the
study of plant breeding. Mendel contributed the law of inde-
pendent inheritance, and Bateson and Punnett in 1906 gave the
first case of linkage in sweet peas in a cross between a purple-
flowered variety with long pollen and a red-flowered variety with
round pollen. The phenotypic condition in the backcross of
50 purple long, 7 purple round, 8 red long, and 47 red round plants
was explained on the basis of gametic production in the ratio of
7 purple long, 1 purple round, 1 red long, and 7 red round Instead
of by the usual gametic ratio of 1:1:1:1. The parental com-
binations were formed seven times as frequently a# the new
combinations,
11
12
METHODS OF PLANT BREEDING
TABLE 1. CHROMOSOME NUMBER (HAPLOID) IN THE COMMON
CROI PLANTS
Scientific name
Common name
Number of
chromosomes
(n)
Cereal Crop Plants and Relatives
Triticum monococcum
Wheat:
ICinkorn
7
Triticum dicoccum
Emmer
14
Triticum durum
Durum
14
Triticum spelta
Speltz
21
Triticum vulgare . .
Bread
21
Avena brevis
Oats:
7
Avena strigosa
7
Avena barbata
14
A vena fatua
Wild
21
Avena sativa
Cultivated
21
Avena byzantina
Red cultivated
21
Avena nuda
Hull-less
21
Hordeum distichon. . ....
Barley :
2-row barley
7
Hordeum deficiens
2-row barley
7
Hordeum vulgare
6-row barley
7
Hordeum jubatum
Squirrel tail
14
Hordeum nodosum
21
Secede cereale
Rve
7
Fagopyrum esculentum
Buckwheat
8
Oryzo, sativa
Rice
12
Zea mays
Corn
10
Sorghum halepensis . . .
Johnson grass
20
Sorghum vulgare
Milo, Kafir,
10
Sorghum vulgare, var. sudanensis
Feterita, Kaoliang
Sudan grass
10
Forage Grasses
Agropyron cristatum
Agropyron pauciflorum. .
Crested wheat
Slender wheat
Red top
Meadow foxtail
Big bluesteni
Little bluestem
Brome grass
Orchard grass
Wild rye
Meadow fescue
Italian rve err ass
7, 14
14
21
14
35
20
21, 28
14
14
7, 14, 21, 35
7
Agrostis alba
Alopecurus pratensis
Andropogon furcatus
Andropogon scoparius
Bromus inermis
Dactylis glomerata
Elymus canadensis
Festuca elatior
Lolium italicum
THE GENETIC AND CYTOGENETIC BASIS
13
TABLE 1, CHROMOSOME NUMBER (HAPLOID) IN THE COMMON
CROP PLANTS. (Continued)
Scientific name
Common name
Number of
chromosomes
(n)
Forage Grasses (Continued)
Lolium perenne Perennial rye grass 7
Panicum miliaceum Proso millet 18, 21
Phalaris arundinaceae Reed canary 7, 14
Phleum pratense Timothy (American) 21
Phleum pratense Timothy (British) 7
Poa compressa Canada blue 7, 21, 28
Poa pratensis Kentucky blue 14-49
Legumes
Glycine soja Soybean 20
Lespedeza sp Japan clover 9, 10, 18
Medicago falcata Alfalfa 8, 16
Medicago sativa Alfalfa 16
Melilotus alba Sweet clover (white) 8
Melilotus officinalis Sweet clover (yellow) 8
Pisum sativum Pea 7
Trifolium hybridum Alsike clover 8
TrifoUum pratense Red clover 7, 12 ^
Trifolium repens White Dutch clover 8, 12, 14, 16
Vigna sinensis Cowpea 12
Fiber Plants
Cannabis sativa Hemp 10
Gossypium sp Cotton (Asiatic) 13
Gossypium sp Cotton (American) 26
Linum usitatissimum Flax 15, 16
Sugar Plants
Beta vulgaris Sugar beet 9
Saccharum offidnarum Sugar cane 40-63
Stimulants
Coffea sp Coffee 11, 22, 33, 44
Nicotiana tabacum Tobacco 24
Thea sinensis Tea 12-13, 15, 22-23
Oil Plants
Aleurites sp Tung oil 11
Arachis hypogaea Peanut 10, 20
Linum usitatissimum Flax 15, 16
Sesamum indicum .... Sesame 26
14
METHODS OF PLANT BREEDING
TABLE 1. CHROMOSOME NUMBER (HAPLOID) IN THE COMMON
CROP PLANTS. (Continued)
Scientific name
Common name
Number of
chromosomes
(n)
Vegetables
Allium cepa
Onion
8
Asparagus officinalis ....
Asparagus
10
Beta vulgaris .
Beet
9
Beta vulgaris var. cicla. . . .
Chard
9
Brassica oleracea
Cabbage, cauliflower,
9
Brassica rapa
kohlrabi
Turnip
10
Capsicum annuum
PeDDer
12
Citrullus vulgaris . . .
Watermelon
11
Cucumis melo
Muskmelon
12
Cucumis sativus
Cucumber
7
Cucurbita moschata
Squash
20
Cucurbita pepo
Pumpkin
20
Lactuca saliva
Lettuce
9
Lycopersicum esculentum
Tomato
12
Phaseolus lunatus
Bean (lima)
11
Phaseolus vulgaris
Bean (kidney)
11
Pisum sativum
Pea
7
Raphanus sativus
Radish
9
Rheum rhaponticum
Rhubarb
22
Solanum melongena
Eggplant
12
Solanuin tuberosum
Potato
24
Svinacia oleracea
Spinach
6
Fruits
Citrus grandis ,
Grapefruit
9
Citrus limonia
Lemon
9
Citrus sinensis
Common orange
9, 18
Fragaria grandijlora
Strawberry (cultivated)
28
Malus malus
Apple
17, 5 H
Prunus americana
Plum (American)
8
Prunus domestica
Plum (European)
24
Prunus avium
Cherry (sweet)
8
Prunus cerasus
Cherry (sour)
16
Prunus persica
Peach
8
Pyrus communis .
Pear
17, 5 Ji
Ribes sp
Currant
8
Rubus idaeus
Red raspberry (Euro-
7, 14
Rubus strigosus
pean)
Red raspberry (Ameri-
7
Yitis sp
can)
Grape (cultivated)
19, 20, 38
THE GENETIC AND CYTOGENETIC BASIS 15
The frequencies of new combinations of factor pairs lying in
homologous chromosomes are dependent to a great extent upon
the distance apart of the genes in the chromosome. The spindle
fiber apparently reduces crossing over in adj acent regions. There
is no interference between crossovers on opposite sides of the
spindle fiber. Many studies of genetic linkages have shown wide
differences between genetic maps and physical-map distances as
determined by cytogenetic study of induced breaks in chromo-
somes. These studies have given added evidence, however, of
the linear order of the genes in the chromosome. There are also
other types of cytological aberrations, such as inversions and
translocations that lead to new genetic maps. In crosses between
the new types with standard types, genetic ratios may be greatly
modified.
Qualitative and quantitative differences in chromosomes have
been extensively studied in recent years, and the field of cyto-
genetics is being constantly developed. Information regarding
the number and nature of chromosome differences is being
obtained rather rapidly for many crop plants. The present
chapter will summarize the usual chromosome numbers in
important crop plants arid illustrate how genetic and cytogenetic
principles are being used by the plant breeder. Chromosome
numbers in the common crop plants, taken largely from the 1936
and 1937 U. S. Department of Agriculture Yearbooks, are given in
Table 1.
POLYPLOIDS IN RELATION TO PLANT BREEDING
A study of chromosome numbers in related species of economic
plants shows many multiple series. A common haploid number
in the Gramineae is 7. Species of Triticunij Avena, and Hordeum
with haploid numbers of 7, 14, and 21 are common and illustrate a
type of variation in polyploids, i.e., multiples of a fundamental
number, often referred to as a euploid series. Aneuploidy, or
variation in chromosome numbers not multiples of a fundamental
number, is frequent in some species. Poa pratensis, for example,
varies from 28 to over 100 somatic chromosomes, Aneuploid
chromosome numbers are of more frequent occurrence in species
that have apomictic development than under conditions of sexual
reproduction.
16
METHODS OF PLANT BREEDING
-There are two main types of euploids, namely, allopolyploids
and autopolyploids. These two types are illustrated in Fig. 2.
In this illustration the autopolyploid is of the autotetraploid type
and has four sets of like or homologous chromosomes. There
may be random pairing or mating between each group of the four
homologous chromosomes. An allopolyploid has chromosome
sets from different sources. In the illustration a haploid set from
doubling
F,hybrid, no pairing
doubling
JUUUQQOO
Autotetraoloid, n?4 Amphidiploid ,n =4
FIG. 2. Schematic diagram of development of autotraploid due to doubling
of chromosome number in the zygote and the production of an arnphidiploid
from a cross between related species due to doubling of chromosome number in
the gametes or zygotes when chromosomes are so unlike that pairing is not
obtained in the Fi cross.
species A is so different from that of species B that pairing does
not take place. Doubling of the chromosome number will result
in an allopolyploid of the amphidiploid type.
One of the best known cases pf triplicate factors in a polyploid
of the amphidiploid type is in ^riticum vulfftire, where there may
be three pairs of factors for brownish red color of the kernel, any
one of the three in a dominant condition leading to the develop-
ment of color. This case was given originally by Nilsson-Ehle as
the basis for the multiple-factor theory of quantitative inherit-
THE GENETIC AND CYTOOENETIC BASIS
17
ance, without a knowledge of the fact that T. vulgare is a
polyploid of the hexaploid type with the amphidiploid type of
chromosome pairing, i.e., contains three sets or genoms of seven
bivalents (7 n ) each, or 42 somatic chromosomes. These three
factor pairs for red kernel color may be designated jRifi, Rzr% f
72sr 3 . An illustration of the mode of inheritance of red vs, color-
less kernels may be given where only two of the three factor pairs
are concerned and the parents and Fi produced red kernels.
Variety A
Parental phenotype ................... red *
Parental genotype .................... .Ri-Ri r^rz T%r%
Fi phenotype ......................... red
Fi genotype ..........................
Variety B
red
Since the r 8 r 8 factor pair is in the homozygous recessive condition,
it will have no effect on kernel color and may be disregarded.
The following summary gives the genotype and kernel color of
jP 2 plants and F 3 breeding behavior :
F,
F*
Genotype
color
Breeding behavior
1 /ki/Li/fc2*fc2
Red
Breeds true for red kernel color
2 RiriRzRz
Red
Breeds true for red kernel color
2 RiRiR%r%
Red
Breeds true for red kernel color
4 RiTiR*tr z
Red
Segregates; 15 plants red kernels: 1 plant colorless
1 RiRiTzTz
Red
Breeds true for red kernel color
2 RiTiTzTz
Red
Segregates; 3 plants red: 1 plant colorless
1 TiTiR^Rz
Red
Breeds true for red kernel color
2 TiTijtltzTz
Red
Segregates; 3 plants red:l plant colorless
1 fiTirzrz
Colorless
Breeds true for colorless kernels
Genotypes RiriR^R^ and RiRiR^r^ are illustrations of poly-*
ploids with genetic segregation that has no definite phenotypic
effect. There is a general relation between intensity of color and
number of dominant factors, but the relation is so indefinite that
the number cannot be estimated by visual inspection.
The segregation in F 2 of 15 plants with red kernel color to 1
with colorless kernels results from crossing two varieties, the one
homozygous for RiRir^r^ and ,the other r.iTiRJR^ Of the 15
plants with red kernel color in Fg, 7 breed true in F for red color,
18 METHODS OF PLANT BREEDING
4 segregate in a 15: 1 ratio, and 4 segregate in a 3: 1 ratio. The
plants with colorless kernels in F 2 breed true for this color in F$.
Contrasted with the foregoing allopolyploid type is the auto-
polyploid, such as is obtained in Datura stramonium resulting
from doubling of the diploid chromosome number. The chromo-
some constitution may consist of four identical sets of chromo-
somes. If we take the case where the diploid was heterozygous
for a single factor pair Dd, the autotetraploid will have the geno-
type DDdd. Such a polyploid has a high frequency of quadri-
valent association, and random chiasma formation occurs among
the four homologous chromosomes.
In an autotetraploid for any dominant factor, there may be a
series of genotypes such as DDDD, DDDd, DDdd, Dddd, dddd,
also written Z>4, Dsd, D^d^ etc., respectively. In an autopoly-
ploid there may be chromosome segregation or random chromatid
segregation. Random chromatid segregation occurs only when
the factors concerned are somewhat more than 50 crossover units
from the spindle-fiber attachment. For closer distances the
ratios are intermediate between those expected from chromosome
and chromatid segregation, approaching chromosome segrega-
tion as the genes become closer to the spindle fiber. This
naturally modifies the ratios obtained from particular types of
heterozygotes.
A convenient method of calculating gametic expectation is
illustrated as follows in an autotetraploid of the D%d (DDDd)
type. The number of combinations of n things taken r at a
n\
time = 7 - rj ,-
(n r)!r!
For chromosome segregation, the gametic expectation may be
calculated in the following manner. Two types of gametes would
be obtained, DD and Dd. The gametic ratio expected is as
follows;
For gamete DD, or the number of different ways of taking two
3!
things out of three, n = 3 and r = 2, and yy^j = 3DD.
For diploid gametes containing a dominant and recessive factor,
Dd, for example, it is unnecessary to use the formula. The D
factor can be taken in three ways from D 8 , whereas d can be
taken in only one. The gametic expectation then will be
3D X Id w Wd. This is the result that would be obtained if the
THE GENETIC AND CYTOGENETIC BASIS 19
formula was used to calculate the number of combinations and if
it is remembered that factorial zero (0!) equals 1.
With random chromatid segregation, however, the condition
would be entirely different. The chromatid condition would be
The number of different ways of obtaining a combination of
DD can be calculated by substituting, where n = 6 and r = 2
in the formula, The frequency of gametes of the DD type will
= 15DD. Gametes of the Dd type
& ' $ * 4: * &
can be obtained by finding how many times one D can be taken
from D 6 , which equals 6D, and this multiplied by 2d would give
l2Dd. One gamete of the dd type can be obtained, and the
gametic ratio then will be 15DD: 12Dd: Idd. Thus, from selfing
DDDd, where the percentage recombination between the D locus
and the spindle fiber approaches 50 per cent, the phenotypic
expectation in Fz is 783D : Id. Such peculiar ratios cannot easily
be differentiated from mutations without studying second-genera-
tion selfed progeny rather extensively.
In a similar manner, the student can calculate expectation for
other genetic types of polyploids. Linkage relations are greatly
complicated in polyploids. The student of plant breeding will be
able to determine logical explanations for the results obtained
only when a knowledge is available of the chromosome mechanism
for the particular plant under study.
There are two general types of euploids, but probably, in many
cases, there are also intermediates that behave as amphidiploids in
some cases and for some chromosome pairs and autopolyploids
under other conditions or for other chromosome pairs. Wheat is
a good example of a hexaploid that generally gives an amphi-
diploid type of inheritance. Ordinary bread wheat contains
three sets, or genoms, of 7 chromosomes each, and, as a rule, the
pairing behavior is of the diploid type. Based upon Winge'p
original explanation, doubling could occur from crosses between
two closely related species each with a chromosome number of
n = 7 } which, through geographical isolation, gene mutations,
and chromosomal changes, had become so differentiated in tfceir
chromosome mechanism that crossing was possible but pairing did
not occur in meiosis. This may have led to the inclusion of
28 chromosomes in a single cell, resulting from equational division
20
METHODS OF PLANT BREEDING
of 14 unpaired chromosomes, 7 being obtained from each parental
species. A further cross between this species with another
closely related form with n = 7 chromosomes would furnish the
type basis for an amphidiploid form with n = 21 chromosomes.
Doubling has occurred in eKperimental material of this nature
both in the zygote and in the gamete.
There are many illustrations in the literature of pairing of the
autopolyploid type in a polyploid that usually pairs as an amphi-
diploid. Breeders of crop plants with the amphidiploid-chromo-
some condition obtain variations in breeding behavior that seem
most logically explained on the basis of changes in the type of
chromosome pairing.
Genom analyses based on types of chromosome pairing have
been made extensively with Triticum and related genera by
numerous investigators. One of the first summaries was that
given by Gaines and Aase (1926), illustrated by Fig. 3.
TRITICUM TURGIDUM TRITICUM VULGARC AEGlLOPS CYLIKDRICA
FIG. 3. Diagram illustrating hypothetical relationships of chromosomes.
The 7 chromosomes in set a and the 7 chromosomes in set 6 are present in both
Triticum vulgare (21 chromosomes as the hapioid number) and in T. turgidum
(14 chromosomes). The 7 chromosomes in set c are present in T. vulgare and
in Avgilops cylindrica but not in Triticum turgidum. The 7 chromosomes in
set d are present in Aegilops but liot in either T. vulgare or T, turgidum. A
21-chromosome wheat X a 14-chromosome wheat gives rise to sporocytes with
14 paired and 7 unpaired chromosomes (lower left). A 21-chromosome wheat
X Aegilops cylindrica gives rise to sporocytes with 7 paired and 21 unpaired
chromosomes (lower center). Aegilops cylindrica X Triticum turgidum gives
rise to sporocytes with 28 unpaired chromosomes (lower right).
The present status of the problem may be summarized as
follows:
THE GENETIC AND CYTOGENET1C BASIS
21
EINKORN SERIES
(n-7)
AA
Triticum aegilopoides
Triticum monococcum
TIMOPHEEVI SERIES
(n - 14)
AAGG
Triticum timopheevi
EMMER SERIES
(n - 14)
AABB
Triticum dicoccoides
Triticum dicoccum
Triticum durum
Triticum turgidum
Triticum pyramidak
Triticum polonicum
SPELT SERIES
(n - 21)
AABBCC
Triticum speUa
Triticum vulgare
Triticum compactum
SECALE SERIES
(n-7)
EE
Secale cereak
AEGILOPS SERIES
(n - 14)
CCDD
Aegilops cylindrica
The species of Triticum, Secale, and Aegilops are seen to be
made up of one or more sets (genoms) of seven chromosomes
each, designated A, J3, C, D, JB, and G (Lilienfeld and Kihara 1934,
Kostoff 1937). In crosses between the Emmer and Spelt series,
for example, the pairing behavior in F\ most commonly consists
of 14n and 7i, although in some cases a few trivalents and quadri-
valents may be obtained because of the fact that chromosomes of
one genom have some homology with those of a different genom.
One of the genoms of timopheevi is similar to the A genom, the
other (GG) resembles B more closely than C but differs rather
widely from jB, forming from two to seven loose conjugations
with it.
Stadler (1928, 1929) studied rate of induced mutation per r unit
in barley, oats, and wheat in relation to chromosome numbers.
Results are as follows:
Species
Number of
chromosomes (n)
Rate of
mutation
Hordeum vulgare .
7
4.9 09
Avend br&vis
7
4.1 1.2
Avena strigosa
14
2.6 0.6
Avena s&tiva
21
Triticum monococcum
7
10.4 3.4
Triticum dicoccum
14
2.0 13
Triticum durum .
14
1 9 5
Triticum vulgare
21
22 ' METHODS OF^ PLANT BREEDING
In general, as has been mentioned, there may be three sets of
factors in hexaploid wheat and oats but o,nly a single factor pair in
each locus for diploid species. A mutation in a homozygous form
A A in barley, giving A a, would produce progeny containing
25 per cent of recessives. In a tetraploid of the amphidiploid
type, two such simultaneous mutations would be necessary in
order than an induced mutation for a character that was doubly
dominant could show up in the immediate progeny.
Variation in pairing whereby one chromosome pair of a genom
shows some homology with a member of a different set would lead
to abnormal segregation. In polyploids of the amphidiploid
type, such results are probably of relatively frequent occurrence.
Powers (1932) and Myers and Powers (1938) have studied
variability in strains of wheat due to various types of chromosome
abnormalities or to gene differences. In a study of Marquillo, a
variety belonging to the spelt series with a haploid chromosome
complement of 21 but derived from a cross between varieties of
Triticum durum and T. vulgare with 14 and 21 haploid chromo-
somes, respectively, Powers (1932) found that germinal instability
in Marquillo was greater than in Marquis or Thatcher. Thatcher
is a variety produced by crossing a sister selection of Marquillo
with a purified hybrid of Marquis X Kanred.
In these studies, the easiest method of estimating the percent-
age of germinal instability was to measure the occurrence of
chromatin loss, measured by the frequency of occurrence
of microspores showing micronuclei. The mean percentage of
micronuclei in four varieties is \given in the following summary,
taken from Myers and Powers:
Variety
Total plants
Mean percentage
of micronuclei
Thatcher
25
8
Marquis ....
26
0.9
H44
20
4.1
Supreme
9
8.3
H44 is a variety of wheat produced by McFadden from a cross
of Yaroslav emmer X Marquis, It has the chromosome number
of the spelt group. Supreme is a variety of T. vulgare produced
by selection from Red Robs.
THE GENETIC AND CYTOGENETIC BASIS 23
Although it seems probable that a wheat of recent origin such
as Marquillo may show greater germinal instability than old
established varieties such as Marquis, as has been pointed out by
Powers and also by Love (1938), it seems of interest that Supreme,
a variety of T. vulgare, selected from a variety not of recent
origin, is equally instable, although there is the possibility that its
origin may be also the result of a natural cross between species.
Myers and Powers showed germinal instability to be inherited,
and the isolation of apparently homozygous lines with different
percentages of micronuclei was considered to indicate that genetic
factors were involved in conditioning meiotic instability.
In the studies of Marquillo, Powers found evidence for 7.2 per
cent of natural crossing. This is higher than has been usually
observed with other varieties of wheat at the Minnesota station.
Thirty-two plants of Marquillo were studied, two of these having
only 41 chromosomes. An average of 23.4 0.24 per cent of the
mierospores of the 41 chromosome plants showed micronuclei,
whereas only 2.8 0.16 per cent of the 42 chromosome plants
showed micronuclei.
In a recent cross in oats by Hayes, Moore, and Stakman (1939)
between Bond, Avena byzantina, and standard varieties of A.
sativa, segregation in F% for type of base on the lower floret
occurred in a 3 : 1 ratio of sativa to byzantina types. Several F 3
families from F% plants showed wide deviations from the type of
segregation in F% and an intermediate type of base bred true in
later generations. The hypothesis that these results were due to
change in chromosome pairing was used, although further studies
are necessary to prove the hypothesis.
Hope and H44 are vulgare wheats with n = 21 chromosomes
that descended from crosses of Triticum dicoccum X Marquis
(T. vulgare). In many studies of stem-rust reaction, when Hope
and H44 are crossed with other varieties of vulgare wheats, wide
deviations from the usual type of F% segregation have been
observed in F$ families. In such crosses, however, it has been
relatively easy to obtain homozygous types with stem-rust
resistance similar to that of the Hope and H44 parents. Changes
in the manner of chromosome pairing in complex polyploids of
the amphidiploid type seem to occur rather frequently. This
tends to complicate breeding behavior and necessitates that
greater care be used to ensure the selections of greatest promise
24 METHODS OF PLANT BREEDING
are breeding true before they are increased for distribution. A
tendency for a modified type of segregation in jP 8 in some families
does not necessarily greatly complicate the breeder's problem of
selecting desirable homozygous types.
Speltoid wheats and fatuoid oats have occurred in T. vulgare
(2n = 42) and A. saliva (2n 42) as a result of chromosomal
variations due very probably to a change in chromosome pairing.
The review of Sansome and Philp (1939) has been used in this
summary. Speltoid wheats resemble T. spelta and fatuoid, or
false wild oats, resemble A. Jatua. Three types have been
observed: Type A, with no change in chromosome numbers,
Type B, with a chromosome deficiency, and Type C, with a
chromosome excess.
The A type, when heterozygous, gives three types of progeny:
homozygous fatuoid or speltoid, heterozygous, and normals in a
ratio of 1:2:1.
An explanation that has been given by Winge and Huskins is
on the basis of a change in pairing due to the similarity of a
chromosome of one genom with that of another. If we designate
the three chromosome pairs concerned, as A, B, and C, one
chromosome belonging to each of the three genoms, and suppose
the B chromosome carries the speltoid factors and C the normal
ABC
factors epistatic to the speltoid, then the normal type
would breed true, as a rule, for absence of speltoid characters.
If one supposes that B has sufficient homology with (7, so that
occasional pairing occurs between B and C, giving rise to
then gametes ABB and A CC would be obtained. If gamete
ABC
ABB mated with ABC, the heterozygous speltoid form
would be obtained. On selfing, three types of progeny would
result normals, -3-575, heterozygous speltoids, -r^ and homo-
ABB
zygous speltoids, -rg|> ia a ratio of 1 : 2 : 1. Such a homozygous
speltoid would give some quadrivalent associations, as was
observed by Winge, whereas the heterozygous speltoid would
show trivalent and univalent associations that were observed
THE GENETIC AND CYTQGEtfETIC BASIS 25
also. The old hypothesis that fatuoids arise through natural
crosses between Avena sativa and A. fatua is seen to be
untenable.
Winge gave formulas for the B and C types of speltoids where
stands for the loss of a chromosome, consisting of the heterozy-
gous type -r77 and two sorts of homozygotes, one -rw with the
loss of a chromosome, and the other .......... : ..... pp-> with the duplication
of the chromosome B.
Huskins studied breeding behavior, variations in pairing, and
chromosome numbers in fatuoids or false wild oats, obtaining
the same sort of results that have been outlined for speltoids.
These types of results have been presented briefly to emphasize
the difficulties of studying genetics in polyploids. The plant
breeder must deal with polyploids in economic plants, and a.
knowledge of the causes of variability may aid greatly in the
breeding program. Selection for types with chromosome pairing
that will give normal disjunction will aid in establishing uniform
breeding strains. Wide deviations from normal types of segrega-
tion may be expected in some progenies. There is, however,
considerable evidence that by selection for germinal stability, in
many cases, the variability resulting from abnormal pairing may
be overcome. Such selection will often be of great economic
importance, since germinal instability often leads to the produc-
tion of undesirable characters. The loss or gain of one or more
chromosomes in polyploids may lead to the production of an
undesirable type of abnormality, such as speltoid wheat or
fatuoid oats.
SOME APPLICATIONS OF GENETICS TO PLANT BREEDING
The value to the plant breeder of a knowledge of the mode of
inheritance of important characters and the application of genetic
principles to methods of breeding will be illustrated by specific
examples. A problem in oat improvement recently investigated
at the Minnesota Experiment Station illustrates the use of
genetics in a practical breeding problem. The parent varieties
and character differences are summarized here,
26 METHODS OF PLANT BREEDING
Anthony, logold, Rainbow Bond
1. Good yield* Fair yield
2. Good-quality grain Excellent quality grain*
3. Fair straw strength Excellent standing ability*
4. Stem-rust resistance* Stem-rust susceptibility
5. Susceptibility to crown rust Crown-rust resistance*
6. Susceptibility to smuts Smut resistance *
7. Sativa type* Byzantina type
* Characters desired in the hybrid.
Frequently certain crosses give a larger proportion of desirable
offspring than others, probably because of the fact that the geno-
type of the one parent supplements that of the other in a more
satisfactory manner, although the reason why certain crosses give
a greater proportion of desirable progeny than others, in many
cases, cannot be placed on a definite genetic basis. These facts
have led the plant breeder to use several crosses for a specific
problem rather than a single cross.
Anthony, logold, and Rainbow were three recommended
varieties of Avena sativa grown by Minnesota farmers. logold,
because of early maturity, is adapted to southern Minnesota;
the midseason varieties Antony and Rainbow usually yield better
than logold in north central and northern Minnesota. Double
Cross A, also crossed with Bond, was a selection from (Minota X
White Russian) X Black Mesdag, homozygous resistant for the
White Russian type of stem-rust reaction and the Black Mesdag
type of resistance to smut. Although not particularly desirable
in type of kernel, it was outstanding in yielding ability.
Anthony, logold, and Rainbow were selected because they
produced good yields of grain, were resistant to stem rust, caused
by Puccinia graminis avenae Eriks. & Henn., and were of the
sativa type. Cultivated varieties of Avena sativa have proved
more desirable in Minnesota. Bond, a variety of A. byzantina, is
highly resistant to crown rust, Puccinia coronata Corda, and to the
smuts prevalent in Minnesota, Ustilago avenae (Pers.) Jens, and
Ustilago levis (Kellermann & Swingle) Magn. Bond excels also
in ability to withstand lodging and in grain quality, producing
plumper kernels with a higher weight per bushel than the recom-
mended varieties of A. sativa.
A summary of the mode of inheritance of the characters will
help to explain the way that genetic principles can be used in a
breeding program.
THE GENETIC AND CYTOGENETIC BASIS 27
MODE OF INHERITANCE OF CHAKACTBRS IN OAT CROSSES
(ANTHONY, IOGOLD, RAINBOW x BOND)
1. Crown rust. F\ resistant; Fg 9 resistant:? susceptible or 3 resistant:!
susceptible.
2. Stem rust. F\ resistant; F 2 3 resistant: 1 susceptible.
3. Smuts. Using a mixture of races of the two smut species, F\ resistant;
^2 segregation 1 to 3 pairs of genes,
4. Sativa vs. byzantina characters.
a. Spikelet disarticulation. F\ sativa base; Fa 3 sativa base: 1 byzantina
base.
h. Floret disjunction. F\ byzantina type; F 2 1 or 2 pairs of factors,
c. Basal hairs. FI sativa type; F 2 3 sativa: 1 byzantina.
5. Yield. Multiple factors.
6. Time of maturity. Multiple factors,
7. Plumpness of grain. Multiple factors.
The purpose of the crosses was to combine in a single variety
the desirable agronomic characters with resistance to three major
oat diseases, stem rust, crown rust, and smuts. In these studies
the pedigree method of breeding was used. It consisted of grow-
ing the segregating generations as progenies so that individual
plant study was possible, each progency consisting of approxi-
mately 50 plants from a single plant of the previous generation.
Selection was continued until homozygous lines were obtained
that appeared desirable. Then the best lines were determined
from replicated yield trials.
Crown Rust. In the cross of Rainbow X Bond, only a single
factor pair was involved. In F 2 in the other crosses, there were
approximately 9 crown-rust-resistant to 7 susceptible plants.
The Bond type of resistance appeared to be due to the com-
plementary action of two factors. Resistance to each physiologic
race to which Bond was resistant seems to be due to the same
genetic factors.
The illustration shown at the top of page 28 is given where
two factor pairs were necessary to explain the results. F% geno-
types and phehotypes and F 3 breeding behavior ar& summarized.
^2 plants resistant to crown rust were selected. On the
average 1 out of 9 may be expected to breed true for resistance in
Fa. Two types of segregating progenies are expected in Fg, one
segregating on a 3 : 1 basis and the other on a 9 : 7 basis.
Progenies breeding true for resistance to crown rust can be
determined by seedling inoculation in the greenhouse. By grow-
ing from 20 to 30 seedlings from each plant selected and by
28
METHODS OF PLANT BREEDING
^ genotype
1 AABB
2AaBB
2AABb
4AaBb
lAAbb
2Aabb
1 aaBB
ZaaBb
I aabb
phenotype
Resistant to crown rust
Resistant to crown rust
Resistant to crown rust
Resistant to crown rust
Susceptible to crown rust
Susceptible to crown rust
Susceptible to crown rust
Susceptible to crown rust
Susceptible to crown rust
breeding behavior
Breeds true for crown-rust resist-
ance
Segregates, 3 resistant : 1 suscepti-
ble
Segregates, 3 resistant : 1 suscepti-
ble
Segregates, 9 resistant : 7 suscepti-
ble
Breeds true for susceptibility
Breeds true for susceptibility
Breeds true for susceptibility
Breeds true for susceptibility
Breeds true for susceptibility
inoculation with crown rust, the progenies from F* to Ff> that are
homozygous for crown-rust reaction can be isolated. These
breed true for resistance under field conditions.
Stem Rust. Stem-rust reaction is handled in the same manner.
By the use of a single factor pair, the results can be illustrated as
follows, where R stands for resistance and r for susceptibility.
logold, Resistant (RR)
Bond, Susceptible (rr)
i, Resistant (Rr)
F z genotype
F% phenotype
F s breeding behavior
IRR
2Rr
Irr
Resistant
Resistant
Susceptible
Breeds true for resistance
Segregates, 3 resistant:! susceptible
Breeds true for susceptibility
As will be discussed in some detail later, many pathogenic
organisms are mixtures of raceff that can be differentiated only by
their manner of reaction on a series of varieties used as differential
hosts. It has been learned that logold and Rainbow are resistant
to races of stem rust 1, 2, 3, 5 and 7, that Anthony and Double
Cross A are resistant to races 1, 2, and 5, whereas Bond is suscep-
tible to all five races. In this case, there is a series of three alleles.
THE GENETIC AND CYTOOENETIC BASIS 29
for resistance and susceptibility to stem rust that may be called
Ri for resistance to five races; #2 for resistance to races 1, 2, and
5; and r for susceptibility to all races.
Crosses of Bond X Anthony or Double Cross A segregate on a
3 : 1 basis in F 2 , and the only two homozygous types that can be
obtained will be those that are resistant and susceptible, respec-
tively, to the three races 1, 2, and 5.
Crosses of Bond with logold and Rainbow segregate also on a
3 : 1 basis, in the presence of inoculum of races 3 and 7, whether
races 1, 2, and 5 are present or absent, and the two homozygous
types that can be obtained will be resistant and susceptible,
respectively, to the five races 1, 2, 3, 5, and 7. A consideration of
these facts will show that, in crosses of Bond with Anthony,
logold, Rainbow, or Double Cross A, infection only with race
I furnishes a satisfactory basis for isolation of the parental type of
resistance.
If Anthony or Double Cross A are crossed with logold or
Rainbow, all offspring are resistant to races 1, 2, and 5, but
segregation in Fz for reaction to races 3 and 7 will occur, giving a
3 : 1 ratio of resistant to susceptible.
From any cross, therefore, between homozygous members of a
multiple allelic series the only homozygous types for the character
difference that can be recovered will be the parental types.
There is agreement between seedling and mature-plant reaction
for stem-rust and seedling studies can be used in the same general
manner for stem-rust reaction as has been outlined for crown rust.
Several races of stem rust can attack the parental varieties
Anthony or logold, but neither of these varieties has been severely
and widely injured by stem rust in farmers' fields under field
conditions since their introduction, and Anthony has been grown
widely for over 10 years.
Reaction to Smuts. It was somewhat difficult to determine
the genetics of smut reaction. Resistance was dominant over
susceptibility, and segregating progenies may be expected to con-
tain fewer susceptible plants than a homozygous susceptible line.
Parent rows were included approximately every 20 rows through-
out the nursery. In crosses of Bond with Anthony, logold, and
Rainbow, the results in F$ indicated a single major-factor pair for
smut reaction. In the crosses between Double Cross A with
Bond, where both parents were resistant to smut, a few highly
30 METHODS OF PLANT BREEDING
susceptible progenies were obtained in F&. From previous
studies, the resistance of Double Cross A was explained on the
basis of two major-factor pairs. The results in the present cross
were explained on the basis that the resistance to smuts of the
Bond parent was independent in inheritance of the two factors
for smut resistance carried by Double Cross A. When all three
factor pairs were recessive, susceptibility resulted.
The methods used in selecting for disease-resistant plants were
based on a knowledge of the mode of inheritance. An epidemic of
crown rust, stem rust, and smut was induced by methods that will
be outlined later. Crown rust appears first, and resistant,
desirable-appearing plants were selected and tagged about 10
days after heading. Stem rust can be determined at maturity.
Plants resistant to stem and crown rust were selected in progenies
that were free from smut, and smut-free plants were selected also
in progenies that had a lower percentage of smut than the suscep-
tible parents. By these methods it was relatively easy to obtain
a large number of progenies resistant to all three diseases.
Sativa vs. Byzantina Characters. Three pairs of contrasting
characters have been used to differentiate byzantina and sativa
oats. These may be illustrated by Fig. 4, in which are shown the
upper and lower florets of Anthony belonging to Avcna sativa and
Bond, a variety of A. byzantina. The characters may be
explained briefly.
1. Spikelet disarticulation has been defined as the separation of
the lower floret of the oat spikelet from the axis of the spikelet.
Three classes were used to describe the segregating generations:
(a) abscission, typical of Bond, at the right of the figure, leaving a
well-defined, deep oval cavity or " sucker mouth" on the face of
the callus on the base of the lemma of the lower grain; (6) dis-
articulation by fracture, leaving a rough, fractured surface with
little or no cavity at the base of the lemma, characteristic of
A. sativa as illustrated by Anthony in the figure; and (c) dis-
articulation by semiabscission, intermediate between a and 6.
2. Basal hairs, conspicuoijs bristles on the base of the lower
floret. The Bond parent had long abundant hairs, and the A.
sativa parents had short hairs that were abundant, few, or absent,
depending on the sativa parent used.
3. Floret disjunction, defined as the method of separation of the
second floret from the first floret.
THE GENETIC AND CYTOGENETIC BASIS 31
The method shown by Bond in Fig. 4, called disjunction by
basifracture, is characterized by the rachilla segment breaking
near its base and remaining firmly attached to the base of the
upper floret, as contrasted with that in Anthony, which has dis-
junction by disarticulation at the apex of the rachilla segment, the
rachilla segment remaining attached to the lower floret. In
hybrids, many plants seemed intermediate and were characterized
by disjunction by heterofracture, the rachilla segment breaking
FIG. 4. At left, tipper and lower florets of Anthony; at right, upper and lower
florets of Bond; showing the characteristics of these varieties.
transversely in the middle portion. The Double Cross A parent
was homozygous for an intermediate type of floret disjunction of
the heterofracture type.
The type of spikelet disarticulation of Anthony and the absence
of long bristles or basal hairs were dominant in f\ y and segregation
for each pair of characters approached 3:1 in F^. There were a
few intermediate types in P\ with fewer long hairs and smaller
base than is characterized by Bond. In general, the type with
Bond base and long basal hairs bred true in F 3 . Progenies were
obtained, also, in F 3 that bred true for the sativa type; others
segregated as in F 2 . Some F 8 lines were obtained that showed
32 METHODS OF PLANT BREEDING
unusual types of segregation in F&. These may be illustrated
by four lines, the progenies of two F 2 plants of the byzantina
type of base and two of the sativa type.
F 8 line
Type of base F 2
Segregation in F 9
1
2
3
4
Byzantina base
Byzantina base
Sativa base
Sativa base
285:1 1:1 S
13B-.17S
2S:23I:SB
11S:9I:9
B == byzantina base, I = intermediate, 8 = sativa,
A large percentage of F% plants of the byzantina type of base
bred true for byzantina base in F 3 , and approximately one-third of
those with a sativa type of base bred true for sativa base in ^3.
By studying the type of breeding behavior in F 3 and later genera-
tions, it was possible to select those that were homozygous for
type of base. Variations in type of segregation very probably
may be correlated with variation in pairing of the chromosomes,
since both sativa and byzantina oats are amphidiploids of the
hexaploid type. A knowledge of genetics aids the breeder in
obtaining pure breeding types and suggests the discard of those
that show unusual types of segregation. After all, deviations
from a 3:1 ratio are fairly common. It seems desirable to
emphasize the probability that two pairs of factors are involved
that are rather closely linked. Using F% data and placing / and B
types of spikelet disarticulation together and separating for basal
hairs on the basis of length of hair gave a dihybrid ratio of sativa
base and short hairs, sativa base and long hairs, byzantina
base and short hairs, and byzantina base and long hairs of
2118:19:81:727, By the product method, this gave a recom-
bination percentage of 2.7 0.3. It seems probable that the
wide deviation from expectation of the two middle classes may
be a result of abnormal chromosomal pairing or other abnormal-
ity. In the absence of long basal hairs, the type of base seemed
to be a little less well developed than where the factors for
byzantina base and those for long basal hairs were both present
in a homozygous condition.
The sativa type of base seems more desirable than the byzan-
tina typQ, because there was a definite tendency for correlation
between shattering and the byzantina base.
THE GENETIC AND CYTOOENETIC BASIS
33
Floret disjunction in the crosses, except where Double Cross A
was one of the parents, was dependent upon at least two pairs of
factors. Double Cross A, when crossed with Bond, gave segrega-
tion that was relatively well explained by a single factor pair,
This factor pair was linked in inheritance with the factor pair for
spikelet disarticulation and also with the factor pair for basal
hairs. A knowledge of inheritance of these three pairs of factors
was an aid in selecting the types that bred true. In this case, the
sativa type was desired with spikelet disarticulation by fracture,
short basal hairs, and sativa type of floret disjunction, the rather
strong linkage aiding in obtaining pure breeding types, since more
of the parental types for all three characters were obtained than
would have been secured in the absence of linkage.
Quantitative Characters. The type of results often obtained
from characters that fluctuate greatly may be illustrated by
plumpness of grain. When sufficient seed is available, weight per
bushel is an easy character to work with. In the small grains, as
studied in Minnesota, there is usually a high correlation between
plumpness of grain and yield, hybrids with well-developed seeds,
relatively free from shriveling, generally yielding much better,
on the average, than those that show a lower degree of plumpness.
Selection during the segregating generations, therefore, is made
for plants with plump seeds; this is done by visual examination.
Behavior in a cross between Double Cross A and Bond is used for
illustration.
Parent variety or F z
Plumpness of grain classes, per cent
0-25
26-50
51-75
76-100
Number of plants in class
Bond
1
5
48
6
28
102
54
26
534
61
381
Double Cross A
F 2
Plumpness of grain is without doubt an inherited character,
dependent upon reaction to diseases and physiological characters
that may influence the metabolism of the plant. By selection
during the segregation generations for freedom from disease and
34 METHODS OF PLANT BREEDING
for plump grain, itVas possible to obtain hybrids that were resist-
ant to all three diseases that were of the sativa type and that had
plumper grain, with higher weight per bushel, than the sativa
parental varieties.
Selection for yield on the individual plant basis seems of little
value, since environmental conditions seem the major cause for
variations. This is shown by the extreme variation in yield per
plant within the parental varieties. All that can be accomplished
during the segregating generations seems to be the selection for
the combination of characters desired. Selection of progenies of
desirable agronomic type seems a desirable practice, with the use
of visual examination rather than intensive study. When
homozygous lines are available, those that yield most satisfac-
torily can be isolated through actual comparative-yield trials.
COLCHICINE AS A POLYPLOIDIZHNTG AGENT
Dermen (1940), in a recent review, has summarized the rather
extensive literature on the methods of producing polyploids
through the use of colchicine. This summary has been used
freely. Heat and cold, X rays and radium, as well as ultraviolet
rays, have been used by various workers to induce chromosomal
aberrations. The discovery that colchicine was a satisfactory
medium for inducing chromosome doubling has made its use
extremely popular and has given the plant breeder a relatively
efficient tcchnic that may be used in the production of polyploid
species and varieties. The technic first became available in 1937.
Although there are 179 literature citations in Dermen's review,
the subject of induced ploidy is of such recent origin that it is
rather difficult accurately to evaluate its practical possibilities.
Dermen quotes Blakeslee (1939) as follows: "We now have an
opportunity to make new species to order/ 7 and " . . . thepossi-
bilities in the way of new forms of economic value seem very
great." He quotes Vavilov (1939), who has said: "The possibili-
ties opened up by the artificial induction of amphidiploidy, i.e.,
of chromosome doubling in hybrids, are immense. Genetics is
entering a new era of extensive "application of distant hybridiza-
tion, at least in the case of plants."
The information already available indicates that polyploids in
horticultural species that can be propagated asexually may be
expected to be of considerable economic value. Autopolyploids
THE GENETIC AND CYTOGENETIC BASIS
35
are frequently larger in size and have more showy flowers than
their diploid ancestors. Emsweller and Brierley (1940) present
results with Lilium formosanum to show the relative ease of
doubling chromosome number. They used 20 one-year-old
plants, trimmed off the tip leaves when the flowering stalk was
6 or 8 in. high, and treated the apical meristem with colchicine
solution for a 2-hr, period. From 3.1. aerial bulblets produced in
FIG. 5. Types of snapdragons. (A) Tetraploid hybrid between tetraploid
Velvet Beauty and tetraploid Red Shades. Note the large ruffled flowers and
very deep color, also the heavier stem. The tetraploid hybrid is setting an
abundance of seed, whereas the parents are highly sterile. (B) Triploid Velvet
Beauty produced by crossing tetraploid Velvet Beauty X diploid Velvet Beauty.
This plant is partially fertile. (C) Diploid Red Shades X Velvet Beauty.
Compared with A, the flowers are smaller, less ruffled, and less deep in color.
(Courtesy of Nebel and Ruttle.}
the axils of the old leaf stubs on the thickened stem apex, 22
polyploids were obtained. They state that polyploids had larger
flowers, pollen grains, and stomata than diploids.
Seeds of crimson flowering tobacco, Nicotiana sandarae, which
has nine pairs of chromosomes, were treated with colchicine, and
autotetraploids were obtained by Warmke and Blakeslee (1939).
The autotetraploids obtained had thicker and broader leaves than
their diploid parents and grew to a larger size and produced larger
36 METHODS OF PLANT BREEDING
and more showy flowers. When octoploids of Nicotiana tabacum
and JV. rustica, which themselves are amphidiploids, were pro-
duced (Smith 1939), the resulting plants showed a general lack
of vigor.
Nebel and Ruttle (1938) have pointed out the value of tetra-
ploid marigolds, petunias, and snapdragons that were developed
by the use of colchicine. In several cases, plants were obtained
that were of sturdier growth and produced larger, sturdier flowers
(see Fig. 5).
In the brief review given here, it is impossible to cover all
phases of the economic importance of induced polyploidy. It is
well known that wide crosses are sometimes possible between
species and genera of grasses, although such crosses, in F^ fre-
quently are highly self-sterile. The development of perennial
grasses of the amphidiploid type, with larger seed size, desirable
for forage and range cover, through crosses between Triticum
and Agropyron species is now being investigated by several
workers.
In a little over a year Blakeslee and coworkers (Blakeslee 1939)
succeeded in doubling the chromosome numbers of 65 different
species and varieties of plants. They report doubling in the
following families : Caryophyllaceae, Chenopodiaceae, Coni-
positae, Cruciferae, Cucurbitaceae, Euphorbiaceae, Malvaceae,
Moraceae, Oxalidaceae, Plantaginaceae, Polemoniacea, Portu-
lacaceae, Solonaceae and Violaceae.
Colchicine occurs in the corm of Colchicum autumnale, which
may contain as much as 0.4 per cent by dry weight. A solution
of 0.4 per cent in water may induce doubling in Datura, and one-
thousandth of this concentration causes doubling in Portulaca.
Colchicine in solution is diffusible into plant tissues, exerting its
effect only on cells undergoing cell division. Colchicine prevents
the formation of the mitotic spindle figure and the development
of the cell wall. Cell division into sister cells is prevented, and
the chromosomes continue to divide. The process of chromo-
some division may continue as long as the tissue is exposed to
colchicine. A summary of the technics of colchicine application
may be of interest.
1. Colchicine in aqueous solution diffuses through plant tissues,
causing internal changes in meristematic tissues as a result of
surface application.
- THE GENETIC AND CYTO&ENETlC BASIS 37
2. Dormant tissues are not affected; only active tissues are
affected by colchicine. Treatment is of value from the practical
standpoint only to tissues that will develop into vegetative,
sexual, or into both types of plant parts.
3. Optimum cultural conditions should be maintained during
treatment so that cell divisfon may be favored.
4. The duration of treatment must be determined for each type
of material. In general, the length of treatment is dependent
upon the time required to complete the cycle of cell division in the
material worked with.
5. Concentration of the colchicine solution should not fall
below an effective minimum and should not be sufficiently high to
be fatal. A satisfactory concentration must be determined for
each material.
Material that has been used with success in colchicine treat-
ment includes seeds, seedlings, growing tips of twigs or buds or
bud scales. Successful applications have been made in the
following media aqueous solution, weak alcohol, a suitable
emulsion, lanolin paste, agar solution, glycerine and water, or
glycerine and alcohol. The range of successful concentration
that has been used varied from 0.0006 per cent to 1 per cent.
Duration of treatment that has proved successful ranges from
merely wetting to 24 hr.
Three methods of treatment that have proved successful will be
outlined briefly.
1. Seed treatment. Seeds of Datura, Cosmos, Portulaca, and
Nicotiana have been soaked in 0.2 to 1.6 per cent aqueous solution
of colchicine. This treatment has been applied successfully to
seeds that will germinate in a few days. Seeds may be planted
after treatment and before germination.
2. Seedling treatment. Germinating seedlings may be im-
mersed in colchicine solution in a shallow container or placed
on filter paper thoroughly wetted with the solution for from 3 to
24 hr. In Cosmos, polyploidy was induced by moistening the
soil over and around the seedlings with a 0.02 to 0.1 per cent
aqueous solution after germination and before the seedlings had
emerged.
3. Treating young shoots or bu$s. Tips of young seedlings of
both woody and herbaceous plants may be treated by brushing
the solution over partially exposed tips once or several times or by
38 METHODS OF PLANT BREEDING
immersing such material in a vessel containing the solution for
the length of time necessary.
Successful treatment has bean obtained by the use of 0.5 to
1.0 per cent colchicine in lanolin smeared on growing portions of
young shoots and on expanding branch buds. Treatment of
young seedlings of flax and petunia has been successful by brush-
ing tepid 1 per cent colchicine agar solution (I part 2 per cent
colchicine to 1 part 3 per cent agar) over the growing tips.
CHAPTER III
MODE OF REPRODUCTION IN RELATION TO
BREEDING METHODS
It is recognized that there is a close relation between mode of
reproduction and methods of breeding. These facts have been
emphasized by Hayes and Garber (1927). Crop plants may be
placed in two groups according to mode of reproduction. These
are (1) asexual and (2) sexual.
THE ASEXUAL GROUP
The most important crop plants belonging to the asexual group
are potatoes, sugar cane, and many fruits. Many of the horticul-
tural plants grown as ornamentals are members of this group also.
Plants belonging to the asexual group are propagated by grafting,
cuttings, layering, or other asexual means. Although this is the
normal method of commercial propagation, reproduction by
sexual means hgfe occurred in asexually propagated varieties or
strains of crop plants at some time in the history of their develop-
ment. Vigor of growth, yielding ability, and other quantitative
characters may be explained genetically, in general, as the result
of the interaction of favorable, partially dominant, growth
factors. With most normal quantitative characters, the number
of these factors is large, and linkage is involved. These factors
account for the reason that it is difficult to obtain all the desirable
growth factors in any one plant in a homozygous condition. If
the more promising plants are selected for propagation, it seems
reasonable to expect these plants to be in a highly heterozygous
condition, and the experience of breeders has shown this to be the*
usual case.
Clonal propagation leads usually to the perpetuation of a
uniform progeny, i.e., to the reproduction of the biotype, but it is
recognized that gene changes or chromosomal aberrations do
occur, although there is some difference of opinion regarding their
frequency. Shamel, Scott, and Pomeroy (1918a,6,c), ia Cali-
39
40 METHODS OF PLANT BREEDING
forma, experimenting with citrus fruits, have based a system of
breeding on selection and propagation from bud sports. The
frequency of bud sports has been emphasized by Shamel and
Pomeroy (1932), who have listed 173 cases of important bud
sports in apples.
Collins and Kerns (1938) discuss mutations in the Cayenne
variety of pineapple. This variety presumably originated as a
vegetatively reproduced progeny of a single plant about 100 years
ago. Thirty mutant types have been shown by progeny trials to
reproduce themselves vegetatively; 8 types have been reproduced
through sexual propagation, and 5 of these proved to be dominant
characters. Collins and Kerns state, "The accumulation of
mutations in asexually propagated forms may conceivably play a ,
role in the running out and acclimatization of varieties. The
parade of agricultural varieties during the past years is a demon-
stration of the chdSiges going on, some of which is known to be
due to progressive or regressive mutations."
Methods of breeding the asexual group may be summarized as
follows:
1. Systematic survey of material.
This is an important step in any breeding program. Such a survey
includes a study of material that is already available and that can be
obtained from any source whatsoever. In most breeding problems, the
wild relatives deserve study also. The various steps in such a survey
may be summarized as follows:
a. Collect and grow a short row, small plot, or several individuals of the
varieties of interest. Classify according to plant characters, both
qualitative and quantitative.
6. Make a systematic study of chromosome numbers and relationships.
c. Study relationships by means of controlled crosses, using both genetic
and cytologic technics.
2. Improvement by clonal selection.
In tree fruits, a careful study of variations that appear as individual
trees, or branches, is of value. A study of the transmission of these
variations must be made by means of a progeny trial. All that is neces-
sary is to compare the performance of selected variations with the normal
variety. This can be accomplished rather quickly by grafting com-
parable trees with the selected variations and with normal budwood and
making a test of the desirability of the two sources of cions. Important
varieties have been selected in the past by these methods. The extent
to which such selection can be made the basis of a standardized breeding
program will depend upon the frequency of such mutations. The selec-
tion of budwood from healthy stock deserves consideration by all who use
this method of propagation as a means of varietal increase.
MODE OF REPRODUCTION 41
In potatoes the tuber-unit or hill-selection method has been used
widely. This method is of value chiefly as a means of keeping the variety
free from degeneration diseases such as the various types of mosaic. It
consists of studying the progeny of selected tubers or hills, selecting the
most desirable clonal lines, and using these as a basis for the commercial
variety. It should be recognized that bud sports do occur occasionally,
and when such are observed that have selection value, they can be used
as a basis for an improved variety. Blodgett and Fernow (1921) origi-
nated the tuber-index method with potatoes as a means of testing for
freedom from disease. The purpose was to test by means of a tuber for
the disease reaction of parent hills and eliminate the diseased hills, the
test being made under greenhouse conditions during the winter months.
This method is now used widely in potato-tuber selection for degeneration
diseases such as mosaic.
3. Breeding plants normally propagated asexually by sexual methods.
Sexual methods of breeding plants belonging to the asexually propa-
gated group are not widely different from those used with other crop
plants. Since asexually propagated varieties are highly heterozygous,
selection in self-fertilized lines is being tried as a means of obtaining
parental varieties with certain desired characters in a homozygous con-
dition. When varietal crosses are used it is of value to determine the
suitability of a particular heterozygous parent variety on the basis of
the characters of its progeny. Crosses between an inbred line that is
relatively homozygous for certain desirable characters, with outstanding
commercial varieties, often furnish the most satisfactory basis for selec-
tion of new and improved asexual varieties.
THE SEXUAL GROUP
Plants belonging to this group may be placed in several subdivi-
sions according to their normal mode of pollination. It should be
recognized that varietal differences of a genotypic nature, as well
as environmental influences, are the major causes of the rather
wide differences that are observed when the normal mode of
pollination of plants of economic importance is studied. The
following subdivisions are those of major importance: naturally
self-pollinated, often cross-pollinated, naturally cross-pollinated,
and dioecious.
Self -pollinated Group.
As a rule, less than 4 per cent of cross-pollination. The
crops generally placed here are barley, wheat, oats, tobacco,
potatoes, flax, rice, peas, beans, soybeans, cowpeas, slender
wheat grass, and tomatoes.
There is a gradual variation in amount of cross-pollination
from this group to that of the often cross-pollinated group and nQ
42
METHODS OF PLANT BREEDING
very clear line of demarcation between the groups. The varia-
tions that occur are a result of either environmental influences,
varietal differences, or a combination of the two causes. As wide
variations in the frequency of natural crosses occur from one
locality to another, it seems unnecessary to summarize the many
detailed studies that have been made. It is important for the
breeder to learn the extent of natural crossing of the crops he is
working with under his own conditions.
Methods of learning the extent of normal cross-pollination are
relatively simple. With tomatoes, at the Connecticut Agricul-
tural Experiment Station, Jones (1916) interplanted alternate
plants of dwarf and standard tomatoes, at the usual spacing, in
rows in the field. Seed from the dwarf plants was harvested and
sown. From 2170 plants that resulted, 43, or approximately
2 per cent, proved to be of standard habit. The extent of natural
cross-pollination would be therefore between 2 and 4 per cent.
Stevenson (1928) studied the extent of cross-pollination in
barley under normal conditions in Minnesota, using Consul and
Gatami as the parental varieties. The type characters and
period of heading of these varieties are as follows:
Variety
Type
character
Date heading
1924
1925
1926
Consul
White
6-27
6-12
6-12
Gatami
Black
6-26
6-12
6-15
Seed of the two varieties was sown alternately in rows spaced
1 ft. apart. Black is dominant over white, and the extent of
natural crossing was determined by sowing seed of the white
glumed variety, collected under the conditions described, and
determining the number of natural crosses. Results from the
3 years are as follows:
Year
White glumed
plants
Black glumed
plants
Per cent off
type
1024
2878
1
0.04
1925
1600
2
0.12
1926
2012
3
0.15
MODS OF REPRODUCTION 43
From similar studies, no natural crosses occurred between
Hanna and Jet, Oderbrucker and Lion, and Manchuria and
Nepal.
Natural crossing has been studied extensively in wheat. There
is a rather wide range in the amount of natural crossing as reported
by investigators located in various parts of the world where
wheat improvement has been carried on, Early investigators,
including DeVries, Biffin, and Fruwirth, considered that natural
crossing was very infrequent. Nilsson-Ehle, in Sweden, stated
that some varieties are cross-pollinated much more frequently
than others. Natural crossing at University Farm, St. Paul,
Minnesota, of at least 2 to 3 per cent, on the average, has been
observed. Powers (1932) studied natural crossing in Marquillo
spring wheat, derived from a cross of lumillo durum with
Marquis. Marquillo was grown in alternate rows with Ceres,
and the percentage of natural crosses determined by inoculation
in the seedling stage with physiologic race 21 of black-stem rust,
Puccinia graminis tritici, to which Marquillo normally is resistant
and Ceres susceptible. Seedlings from seed produced on non-
covered spikes of Marquillo showed 3.6 0.50 per cent of
susceptible plants. Since susceptibility is dominant over resist-
ance to form 21 in crosses of Ceres X Marquillo, it is fair to
conclude that natural crossing to the extent of 7.2 per cent
occurred in Marquillo wheat during the year that the study was
conducted.
K Often Cross-pollinated Group.
In this group, self-pollination is more frequent, as a rule,
than cross-pollination, although cross-pollination may occur
so frequently that some method of preventing cross-pollina-
tion between varieties and strains of different genotypic
constitution must be followed throughout the breeding and
seed-distribution program. Crops belonging to this group
are cotton, sorghums, and some strains of sweet clover.
Except for the necessity of controlling pollination in seed plots
to a greater extent than with the self-pollinated group, it seems
probable that methods of breeding are not greatly different than
for the self -pollinated group.
Before starting a hybridization program of improvement, it
may be desirable to practice self-pollination and selection in order
44 METHODS OF PLANT BREEDING
to isolate the more desirable hoinozygous types as parents and
eliminate the less desirable variations.
f. Naturally Cross-pollinated Group.
Important crop plants placed in this group include maize,
rye, clovers, sunflowers, sugar beets, many fruits, some
annual and most perennial grasses, cucurbits, Brassica
species, most root vegetables.
This group is composed of plants of widely different habit in
relation to mode of pollination. It includes such plants as maize,
with which cross-fertilization is the rule and which sets seed freely
when artificial self-pollination is practiced. The wind-pollina-
tion habit and the large amount of pollen produced tend to cause
cross-pollination that approaches 100 per cent. Then there are
many plants adapted to insect pollination, in which cross-pollina-
tion, under normal conditions, is essential to seed production, and
many plants that are partially or wholly self-incompatible, in
which case cross-pollination is essential to seed production
because of self-sterility.
Self-sterility and other causes of self-unfruitfulness will be
discussed in much greater detail in connection with the presenta-
tion of methods of breeding crop plants that normally do not set
seed by self-pollination.
It is apparent that many crop plants contain genotypes that
carry factors both for self-fertility and sterility. Where self-
sterility is the rule, methods of breeding are not widely different
than in dioecious plants, since two parent plants must be selected
in order to obtain a progeny.
L Dioecious Plants.
Important crop plants of this group are hops, hemp, date
palm, spinach, and asparagus.
In breeding plants belonging to this group, it is necessary to
select both male and female plants with the characters desired
and test their progeny to determine the breeding value of particu-
.ar parents. By this means varieties of superior type may be
synthesized.
SELF-POLLINATION LEADS TO HOMOZYGOSIS
Even though only an occasional natural cross occurs in a
lormally self-pollinated crop, this may lead to a new combination
MODE OF REPRODUCTION 45
of characters and thus be a source of material for selection. It
will be of interest to show what will happen in later generations of
self-fertilization as a result of a cross between varieties differing
by one or more genetic factor pairs. Two somewhat different
formulas have been used to express the expectations (East &
Jones 1919).
Suppose that the two parent varieties differ by several factor
pairs. The following formula, [1 + (2 r l)] w , may be used
where r equals the number of segregating generations after a cross
and n equals the number of independently inherited factor pairs
involved and the first and second terms of the binomial are 1 and
2 r 1, respectively. The exponent of the first term gives the
number of heterozygous factor pairs and the exponent of the
second term, the number of homozygous factor pairs. Supposing
the number of factor pairs is 3, i.e.j n = 3, and the progeny
is in the fifth segregating generation, or F$, i.e., r = 5 and
2' - 1 == 31. The results will be I 8 + 3(1) 2 31 + 3(1)(31) 2 +
31 s , giving:
1 individual with all three factor pairs heterozygous.
93 individuals with two factor pairs heterozygous and one
homozygous.
2883 individuals with one factor pair heterozygous and two
homozygous.
29,791 individuals with all three factor pairs homozygous.
Another formula that has been used to express the percentage
of homozygous individuals in any generation following a cross
between different forms is ( ^ J ; where n and r have the
same meaning as in the previous formula. In actual practice, the
calculated expectation would not hold unless all the progeny of
each genotype were equally productive and the factor pairs were
independently inherited. If linkage is involved, this changes the
percentage of homozygous individuals but does not change the
percentage of homozygosis, as has been shown by Wright (1921).
The percentage of homozygosis in any segregating generation, r,
can be obtained by the foregoing formula for a single factor pair.
Under conditions of self-pollination, linkage increases the rapidity
of obtaining homozygous individuals over that expected for
independent Mendelian inheritance.
The results of applying this formula, with 1, 5, 10, and 15 factor
pairs for from 1 to 10 generations of self-fertilization have been
46
METHODS OF PLANT BREEDING
expressed in the form of curves by Jones (1918). The results
(Fig. 6) are given on the basis of the percentage of heterozygous
individuals in each selfed generation and the percentage of
heterozygous pairs, i.e.j the percentage of heterozygosis.
These graphs show that self-fertilization leads rapidly to homo-
zygosis and that the progeny of individual plants of a self-
fertilized crop may be expected to breed true for the most part.
100 *
Percentage of Heterozygous
Individuals in each Selfed
Generation when the Number
of Alleles Concerned
Are: 1,5 10,15.
1
8
9 10
3456
Segreating Generations
FIG. 6. The percentage of heterozygous individuals in each selfed generation
when the number of independently inherited factor pairs are 1, 5, 10, and 15.
The percentage of heterozygosis in any selfed generation is given by the curve for
one factor pair.
The principles outlined show why breeding methods have been
standardized to a considerable extent with self-pollinated crop
plants.
THE EFFECTS OF SELF POLLINATION IN THE OFTEN
CROSS-POLLINATED GROUP
It has been stated that except for the necessity of greater
care in controlling pollination the breeding of the often cross^
pollinated group can be carried on in much the same manner as
with plants belonging to the naturally self-pollinated group.
MODE OF REPRODUCTION
47
Kearney (1923), with Pima cotton, studied the effects of
controlled self-fertilization during successive generations. Some
of the results presented by him are summarized here.
TABLE 2. RANDOM SAMPLE OF COMMEBCIAL STOCK OF PIMA COTTON
COMPARED WITH STOCK INBRED FOR SEVEN SUCCESSIVE GENERATIONS
Ivtean
Population
Flowers
tagged
Percentage
of bolls
shed
number of
seeds
matured
Mean
weight of
1000 seed,
Percentage
of germi-
nation of
g-
seeds
per boll
Inbred
296
11.8 1.3
17.2 + 0.12
13.6 0.04
90.8 + 0.8
Open pollinated
367
8.4 1.0
17.1 0.12
13.4 0.03
90.2 0.9
Difference . .
3 4 1.6
0.1 17
0.2 0.05
0.6 1.2
Boll Weight and Lint Index
Population
Number of bolls
Seed cotton
Lint index
Inbred
105
3.22 + 0.21
4.90 0.27
Open pollinated
115
3 04 + 06
5 12 03
Difference
0.18 4- 0.22
0.22 27
Boll Dimensions
Population
Number of bolls
Length, mm.
Diameter, mm.
Inbred
25
46.6 4- 56
26 8 -f 19
Open pollinated
25
45 7 + 80
26 1 19
Difference
9 + 97
7 27
There was no harmful effect of continued self-pollination in this
variety of cotton. It may be concluded that controlled self-
pollination can be used, when desired, with plants belonging to
this group without leading to a great reduction in vigor.
Humphrey (1940) has emphasized the desirability of inbreeding
cotton in order to obtain uniformity in fiber characters. Com-
parison of lines that had been self -pollinated for 2 and 7 years
indicated that inbred lines were much more uniform than the
variety from which they arose, but little increase in uniformity
was obtained after 2 years of self-pollination. Humphrey's data
lead to the conclusion that vigorous self -pollinated lines can be
48
METHODS OF PLANT BREEDING
obtained in cotton, and, as would be expected, there seem to be no
harmful effects of continued self-pollination.
EFFECTS OF SELF-FERTILIZATION
IN CROSS-POLLINATED PLANTS
From the genetic standpoint, artificial self-pollination in a
normally cross-pollinated crop leads to the production of homo-
zygous lines. In many crops, notably corn, there is a rapid reduc-
tion in vigor when self-pollination is practiced. The extent that
vigor of growth is reduced is not the same in all lines, and some
inbred lines of corn have been obtained that appear relatively
homozygous and are rather vigorous. In general, in corn, no
inbred lines that approach homozygosis have been obtained that
are as vigorous as normal corn. Studies of the effects of self-
fertilization have been made with many crop plants. Extensive
studies of squashes have been made, and much of the improve-
ment in recent years has resulted from the isolation of desirable
selfed lines and their use as commercial varieties. Both high-
yielding and low-yielding selfed lines have been isolated. Cum-
mings and Jenkins (1928) studied a high-yielding line that had
been selfed for 10 generations without harmful effect.
It is apparent that the extent to which a crop can be inbred
without leading to a great reduction in vigor will be the main
factor in deciding how extensively controlled self-pollination can
TABLE 3. THE EFFECT OF 30 GENERATIONS OF SELF-FERTILIZATION WITH
THREE INBRED LINES OF MAIZE UPON THE HEIGHT OF PLANT AND
YIELD OF GRAIN
Line 1-6
Line 1-7
Line 1-9
Number of
generations
selfed
Height,
in.
Yield,
bu. per
Height,
in.
Yield,
bu. per
Height,
in.
Yield,
bu. per
acre
acre
acre
117
81 7
117
81 7
117
81 7
1-5
87
64 11
81
51+7
77
41+5
6-10
97 1*
45 12
84 1
36+5
82 2
34 4
11-15
97 3
38+4
84 + 2
34 3
83 + 2
26 2
16-20
88+4
22+4
85 + 3
24 3
75 4
14 3
21-25
81 2
20 6
75 3
21 3
71 3
13 2
26-30
92 3
24 9
80 2
18 4
77 3
9 4
* Standard errors.
MODE OF REPRODUCTION
49
be used in breeding cross-pollinated plants. Studies of controlled
cross- and self-pollination with each of the important crop plants
are essential in the establishment of breeding methods.
Jones (1939) has summarized the effects of continued inbreed-
ing with maize for three inbred lines started by East in 1905 and
discussed by East and Hayes (1912). Yield in bushels per acre
and height of plant in inches given in Table 3 were presented by
Jones (1939). The data were given as averages for 5-year
periods to overcome seasonal fluctuations.
I2C
2.5
22.5
27.5
J.5 12,5 17.5
Number of generations sefffed
FIG. 7. A comparison of three maize lines, derived from the same variety,
self-fertilized for 30 generations. Height of stalk is measured in inches and yield
of grain in bushels per acre, both plotted on the same scale. The broken lines
are the theoretical curves of inbreeding. (Adapted from Jones.)
The results are presented also in the form of curves in Fig. 7.
The theoretical curves were calculated by subtracting the
average height and average yield of the three inbred lines at the
end of the 30 generations of self-pollination from the figures at
the start. The difference was halved in each generation and
subtracted from the initial yield. Theoretical yields at the end
of the fifth generation of inbreeding were obtained by averaging
the theoretical yields at the end of each generation from 1 to 5,
with the use of the following calculations, where the original yield
was 81 bu. and the average yield at the end of 30 years of selfing
was an average of 24, 18, and 9, or 17 bu. Subtracting from 81,
50
METHODS OF PLANT BREEDING
one obtains 81 17 = 64. This value is halved for each succes-
sive generation of selfing and subtracted from 81. On this basis,
theoretical yields for each of the first five generations of selfing
can be computed as follows:
Generation
of selfing
Calculation
Theoretical
yield
1
81 - (}i X 64)
49
2
81 - (% X 64)
33
3
81 - (H X 64)
25
4
81 - OJie X 64)
21
5
81 - (% X 64)
19
Average . . .
29.4
i
The calculations are based on the hypothesis that the effects
of inbreeding are dependent upon the extent of heterozygosity,
the number of heterozygous pairs of factors being reduced one-
half for each successive generation of self-fertilization. The
average theoretical yield of the three inbred lines, for generations
1 to 5, of 29.4 bu., was less than the actual yield obtained, which
indicates that selection was practiced.
Comparing the actual and theoretical curves indicates that the
three inbred lines were homozygous for factors influencing height
after 5 generation of selfing and for yield after approximately
20 generations of selfing. Besides these lines that can be propa-
gated by continued self-pollination, there are other inbred strains
that are so weak that they cannot be propagated and still others
that can be perpetuated only with difficulty.
HETEROSIS AND ITS EXPLANATION
Early plant hybridists, including Kolreuter, in the eighteenth
century, Gartner and Weigmann, in the nineteenth, noted the
increased vigor of hybrids. Although many others in the last
century observed hybrid vigor, a clear understanding of the
effects of self-fertilization in cross-pollinated plants and of the
effects of crossing self-pollinated plants was obtained only as a
result of genetic research. East and Hayes (1912) pointed out
the value of hybrid vigor both in evolution and in plant breeding
and Shull (1914) suggested the term heterosis in the following
words: "
MODE OF REPRODUCTION 51
To avoid the implication that all the genotypic differences which
stimulate cell division, growth and other physiological causes are
Mendelian in their inheritance and also to gain brevity of expression, I
suggest that instead of the phrases, "stimulus of heterozygosis,"
" heterozygotic stimulation," . . . , that the word heterosi^ be adopted.
Because of the importance to the plant breeder of heterosis, it
seems desirable to summarize the suggestions for an explanation
of heterosis in the light of present-day knowledge.
Keeble and Pellew (1910) explained hybrid vigor in peas from
a cross between two half-dwarf varieties, Autocrat and Bountiful,
on a dihybrid basis with dominance in the FI of the thick stem of
one parent and long internode of the other.
This explanation was not considered generally applicable by
East and Shull, since other cases of hybrid vigor could not be
placed on an equally simple basis. At this time (around 1910)
there was a lack of appreciation by most workers that for many
characters large numbers of factor pairs were involved. Brief
quotations from East and Hayes (1912) serve to summarize the
viewpoints of East and Shull, who, in general, were in close
agreement.
One can say that greater development stimulus is evolved when the
mate of an allelomorphic pair is lacking than when both are present in
the zygote. In other words, development stimulus is less when like
genes are received from both parents.
The decrease in vigor due to inbreeding naturally cross-fertilized
species and the increase in vigor due to crossing of naturally self-
fertilized species are manifestations of the same phenomenon. This
phenomenon is heterozygosis. Crossing produces heterozygosis in all
characters by which the parent plants differ. Inbreeding tends to
produce homozygosis automatically.
Inbreeding is not injurious in itself but weak types kept in existence
in a cross-fertilized species through heterozygosis may be isolated by this
means. Weak types appear in self-fertilized species, but are eliminated
because they must stand or fall on their own merits.
The selfed lines of corn first grown by East at the Connecticut
station in 1906 and carried on by Hayes from 1910 to 1914 were a
part of the long-time selfed material used by Jones (1918) in
studies of selfing and crossing.
Studies of size inheritance in tobacco and corn furnished a
considerable part of the necessary evidence that many quantita-
52 METHODS OF PLANT BREEDING
tive characters were dependent upon the interaction of many
factors for their full expression. If such is the case, it is evident
from any particular cross that it is difficult to obtain all factors in
a homozygous condition that have an influence on hybrid vigor.
The explanation of heterosis given by Jones (1917) is well known
to all students of genetics. He supposed that the vigor of an FI
cross was dependent upon the interaction of dominant, favorable
growth factors, part of which were obtained from each of the two
parents. If large numbers of factors are involved, linkage is
bound to occur, and this makes it extremely difficult to obtain all
necessary growth factors in a homozygous dominant condition in
later segregating generations.
To the breeder who has had much experience with quantitative
inheritance the explanation seems entirely logical. It is difficult
to prove the truth or falsity of the explanation. To the breeder
of economic crop plants the dominance of linked growth factors in
relation to heterosis furnishes a basis for methods of breeding
cross-pollinated plants and the perpetuation of hybrid vigor to
the extent possible with each particular category of crop plant.
Collins (1921) raised certain objections to Jones's explanation.
He emphasized the importance of deleterious recessives that so
frequently show up as a result of inbreeding maize. He pointed
out also that the effect of a genetic factor was dependent upon the
size of an organism and that skewness of the P\ distributions
would not be evident if as many as 20 pairs of factors were
involved, with complete dominance of each and a cumulative
effect of one on the other. With more pairs of factors involved,
linkage would result, however, and add to the difficulty.
Richey (1927) and Richey and Sprague (1931) have presented
data on convergent improvement that gives some support to
Jones's explanation of hybrid vigor. The method of convergent
improvement will be presented in much greater detail in relation
to corn breeding. It is equivalent to double backcrossing and
furnishes a method for improving each of two inbred lines without
interfering with their combining ability. If the selfed lines A and
B combine to give a vigorous FI cross, (A X -B), two series of
backcrosses are carried on, (A X B}A and (A X B)B. In the
cross of (A X B) A and subsequent backcrosses to A, it is hoped
to retain the favorable dominant growth factors from A and add a
part of B. In the cross of (A X B)B y etc., the favorable domi-
MODE OF REPRODUCTION 58
nant growth factors of B will be retained, and a part of these
from A will be added. After backcrossing, selfing is necessary
until the heterozygous dominant growth factors become homo-
zygous. The Fi cross of A(/?i), containing the growth factors of
A with a part of those obtained from B, with B(Ai) 7 or [A(Bi) X
B(Ai)], should yield as much as A X B if the partially dominant
linked-growth-factor theory of hybrid vigor is the correct one.
Greater yields seem possible if dominance is not complete.
Richey and Sprague (1931) presented data from six crosses
where N = nonrecurring parent, R = recurring parent, and
(N X R*) refers to four generations of backcrossing of (N X R) X
R. All combinations of new crosses [A(Bi) X B(Ai)] should not
be expected to yield equally as well as A X B, and in practice it is
necessary to determine the number of generations to backcross
before selfing. One of the crosses studied by Richey and Sprague
indicates the possibility of increasing the yield of selfed lines and
of the Fi cross over the original selfed lines and FI cross, respec-
tively, by the process of convergent improvement. In a repli-
cated trial, N X R yielded 17.8 0.20 Ib. of ear corn per plot,
N X (N X R*), an Fi cross of a fourth generation backcross of
(N X K)R 9 when crossed with N, yielded 19.0 0.30, or
significantly more than N X R. The yield of R selfed was
5.5 0.22, which is significantly lower than the yield of
(N X RA} of 8.3 0.24.
The writers say, " Convergent improvement, suggested
originally from theoretical considerations as a means of improving
selfed lines of corn without interfering with their behavior in
hybrid combination, so far has been found successful. Further-
more, the results suggest that this method may also provide a
means by which the yields of F\ crosses between selfed lines can
be raised to an even higher level."
More recently East (1936?;) has presented a genetic explanation
for heterosis that emphasizes the importance of linkage and
makes the suggestion that multiple alleles are concerned also in
heterosis. The genetic factors involved are not those used
normally in genetic experiments, called physiologic defectives
by East, but the factors with small effects and more difficult to
study are considered to be of greater importance in evolution and
in plant breeding. It is suggested by East that these factors,
which have a cumulative effect and for which dominance is
54 METHODS OF PLANT BREEDING
virtually absent, occur in series of multiple alleles. Each mem-
ber of a series may be considered to have the ability of affecting
a different physiological process. Thus, if A\, A%, and A$ are
three such alleles, AiA% or any other combination of two of the
three factors would have a greater effect than the homozygous
condition for one, i.e., A\Ai, A 2^2, or A^A^. Although there is
nothing inconsistent in such a hypothesis in the light of the
numerous series of multiple alleles for qualitative characters, there
seems no reason to suppose that multiple alleles are of greater
importance for quantitative than for qualitative characters.
Studies of corn breeding have given further evidence regarding
hybrid vigor, although the problem of heterosis needs further
study, both on a genetic and physiological basis. It is now
generally accepted by students of corn breeding that combining
ability is a genetic character. Recent rather extensive studies
of Hayes and Johnson (1939) showed the extent to which com-
bining ability is an inherited character. When selfed lines were
selected from a cross between inbreds with high combining ability,
most of the selfed lines obtained from the cross were of high com-
bining ability also, as tested in inbred-variety crosses. Con-
versely, when selfed lines were selected from a cross of low
combiners, the greater proportion of lines obtained were of low
combining ability. Data were given to show that the characters
of selfed lines that measure vigor of growth were responsible for
approximately 45 per cent of the variance in yield of the inbred-
variety crosses.
Heterosis, then, is a general term for hybrid vigor. It is a
phase of quantitative inheritance, and if quantitative inheritance
is Mendelian it seems equally reasonable to place heterosis in a
similar category. If the growth characters of self-pollinated
plants are inherited in the same manner as in cross-pollinated
plants, it seems evident that nature and man have obtained
vigorous self-pollinated plants by selection of the fittest. It is
evident that similar selection in selfed lines of cross-pollinated
plants will, in many cases, lead to the isolation of inbred lines
that are progressively more vigorous than those now available.
The extent of improvement obtainable can be determined only
by actual study.
In recent years, physiological studies of the manifestations of
heterosis have been made with several different crops by Ashby
MODE OF REPRODUCTION
(1930, -1932, 1937), Sprague (1936), Lindstrom (1935), Luckwill
(1937), et al. In general, three stages of development may be
differentiated: (1) fertilization to maturity of seed; (2) from
germination to first flowering; (3) subsequent growth. The
efficiency of the FI crosses was studied for various physiological
characters. The hybrids during the stages of development
designated as (2) and (3) did not excel the better inbred
parent in relative growth rate, respiration rate, or assimilation
rate. Ashby attributes the greater development of the hybrid
to "greater initial capital/' i.e., greater embryo size. Although
Ashby's data were in agreement with this hypothesis, no such
relation is universally present. Sprague (1936) concluded that
growth rate of the hybrids was greater than that of the inbreds
during the first stage and in the early seedling stage but could
not demonstrate a higher growth rate for the hybrids from the
late seedling stage to maturity.
Kiesselbach (1922) studied external and internal expressions
of hybrid vigor in maize crosses. The increased weight of kernel
due to crossing showed the following percentage of increase of
parts of the hybrid kernel over the kernels of the inbreds: total
kernel, 11.1 per cent; embryo, 20.2 per cent; endosperm, 10.4
per cent; and seed coat, 5.4 per cent.
Some measurements of the causes of increased vigor given by
Kiesselbach are of interest.
Increase of hybrids over their pure-line parents:
Stalk diameter at base, 48 per cent.
Number of fibro-vascular bundles in cross section of stalk,
43 per cent.
Number of fibro-vascular bundles in 1 sq. cm. of cross section,
38 per cent.
Average diameter of one pith cell in stalk, 6 per cent.
Average length of one pith cell in stalk, 10 per cent.
Number of pith cells along one diameter in cross section,
38 per cent.
Increase in size of the hybrid over the parents in pith cells in
the stalk and epidermal cells of the leaf was studied in relation to
cell number and cell size. The total increase of the hybrid over
its parents was due to 10.6 per cent increase in cell size and 89.4
per cent to an increase in cell number.
56 METHODS OF PLANT BREEDING
Bindloss (1938), Whaley (1939a,6), and Wang (1939) studied
the apical meristem of inbreds and FI hybrids without finding
any one characteristic uniformly correlated with hybrid vigor.
Bindloss observed a positive correlation between nuclear size
and heterosis in one maize pedigree but no such relation in two
others studied. Her data indicate significantly larger nuclei for
the hybrid than for either parent in one cross, but in another
hybrid the nuclei in the meristem were intermediate between the
two inbred parents. Whaley found that cell and nuclear size
in the plumular meristem of Lycopersicum decreases during
development but less rapidly in the hybrids than in their parents.
The differences observed indicate a fundamental metabolic dif-
ference between the hybrids and their parents. Wang studied
four inbred lines of corn and all six possible F\ crosses between
them, using the apical meristem of the growing shoot. He found
some evidence of heterosis in the volume of the plumular meri-
stem and within the hybrids or within the selfed lines a positive
correlation between cytonuclear ratio of the cells of the growing
shoot and vigor of growth. This ratio, however, did not hold
when comparisons of hybrids and selfed lines were made.
From these physiological studies, there is an indication that
the hybrid approaches the better parent in measures of physio-
logic efficiency. The lack of agreement among the various
studies indicates that heterosis is manifested in various ways in
different hybrids and that it may be due to various causes. The
hypothesis of the complementary action of growth genes seems
the best genetic explanation now available. For the plant
breeder, the explanation of Jones for heterosis on the basis of the
partial dominance of linked growth factors furnishes, at any
rate, a working basis that aids in an attack on improvement
problems. Considering heterosis as a phase of quantitative
inheritance furnishes a basis for an outline of methods of breeding
that aim to obtain, as far as possible, the full benefits of hybrid
vigor to the grower and producer of crop plants.
A CLASSIFICATION OF METHODS OF BREEDING SEXUALLY
PROPAGATED PLANTS
A brief outline of methods of breeding will help to illustrate
the close relation between methods of breeding and mode of pol-
lination, The major groups are as follows:
MODE OF REPRODUCTION 57
I. Introductions.
II. Selections.
A. Mass selection.
1. In self-pollinated crops.
2. In cross-pollinated crops.
3. In dioecious crops. Selection of both male and female plants for
the characters desired.
B, Individual plant selection.
1. In self-pollinated crops.
2. In cross-pollinated crops without control of pollination.
3. In controlled self-pollinated lines of cross-pollinated plants.
4. In dioecious crops.
5. In crops normally clonally propagated.
III. Hybridization.
A . Crosses in self-pollinated e> ops.
1. The pedigree and bulk methods.
2. Backcrossing.
B. Crosses of self-pollinated lines and the use of the FI generation for
the commercial crop.
C. Convergent improvement.
These various methods will be outlined in greater detail later.
At this time it will be sufficient to discuss them briefly.
Introduction is not a method of breeding in itself but a means
of securing material from other workers and from foreign coun-
tries. Many species and varieties of crop plants now grown in
one country were introduced originally from foreign countries.
For example, the soybean, introduced into the United States
from the Orient in the present century, is becoming of out-
standing value to American agriculture.
Mass selection as now practiced in self -pollinated crops is
chiefly a matter of roguing or of selecting individual plants or
heads from a commercial standard variety for seed-plot pur-
poses. In cross-pollinated plants, mass selection is of great value
as a means of selecting and developing ecotypes that through
years of natural selection have become adapted to particular
environmental conditions. Grimm alfalfa, selected in Carver
County, Minnesota, many years ago, was a product of mass
selection.
More varieties of self-pollinated crops have been obtained
from the individual-plant method of selection than by other
methods. Some of the results of these and other methods of
breeding have been summarized by Hunter and Leake (1933).
Most commercial varieties are mixtures of different biotypes that
58 METHODS OF PLANT BREEDING
can be isolated by the individual-plant-selection method. These
mixtures result from natural crossing, mutation, or from mechani-
cal mixtures. They have furnished a logical basis for the
selection of pure-line strains of greatest promise. Several
varieties of oats, Gold Rain and Victory at Svalof, Sweden,
Gopher in Minnesota, Richland, lowar, and logold from Iowa,
and Rusota from North Dakota are illustrations of valuable
varieties obtained by this method of breeding.
With cross-pollinated crops one of the best known illustrations
of individual-plant methods of selection is the ear-to-row-
selection method with corn outlined by Hopkins about 1900.
This method has been used rather widely as a means of develop-
ing adapted varieties of corn. Most of the improvement in
sugar content and quality, of a heritable nature, with sugar
beets was a result of individual-plant selection without control
of pollination.
With cross-pollinated plants like corn, selection in self-
fertilized lines has been used during the last 15-year period as
one of the steps in the modern corn-breeding program. It has
been used also with potatoes as a means of developing better
breeding stock.
With dioecious plants, a good illustration of the individual-
plant methods of selection that led to the development of an
improved variety is the Washington asparagus listed in many
seed catalogues. In this case, both male and female parent
plants were selected and their combining ability determined.
Hybridization is a means of combining the desirable char-
acters of two or more varieties. Two smooth-awn varieties of
barley Velvet, developed at the Minnesota Agricultural Experi-
ment Station, and Barbless, in Wisconsin and the Little Joss
and Yeoman varieties of wheat developed in England are illus-
trations of the many cases in recent years where new varieties of
crop plants have been developed by combining in a single
variety the desirable characters of two or more parents.
The development of a hybrid method of seed-corn production
was predicted by Shull in 1909. Some of the results of this
method of breeding have been emphasized in the first chapter.
As a result of combined genetic and plant-breeding studies,
the value of backcrossing in plant breeding is becoming generally
recognized. When it is desired to add one or two characters to
MODE OF REPRODUCTION 59
an otherwise desirable variety and the technic of crossing is
relatively easy, the method seems almost to be made to order.
Convergent improvement or double backcrossing is a method
of improving each of two inbred lines of corn or other crop plant
without modifying their combining ability.
These and other methods of plant breeding will be discussed in
greater detail in later chapters, when the relative desirability of
various methods of breeding for different types of improvement
problems will be emphasized.
CHAPTER IV
TECHNICS IN SELFING AND CROSSING
Two general methods for the exclusion of foreign pollen may
be used in controlled self-pollination. One method is the use of
space isolation, spacing single plants far enough apart from other
plants with which they might cross so that selfing is ensured.
The distance needed for complete isolation will vary with the
crop, weather conditions, and natural barriers to the spread of
pollen. This method has been employed rather extensively in
selfing sugar beets. The other method is the use of some type
of bag, either paper, vegetable parchment, or cloth, to enclose
the inflorescences and ensure self-pollination.
Crossing different strains usually involves the use of some
special technic appropriate for the crop and environmental con-
ditions prevailing. A knowledge of flower structure of the
species or variety to be worked with is essential before crossing
is undertaken. Some important features of the technic of cross-
ing have been summarized by Hayes and Garber (1927) as
follows:
/ 1. Make a careful study of the structure of the flower before
commencing operations. This may be with, or without, the
aid of a dissecting microscope.
/ 2. Determine which flowers produce the larger, healthier
seeds and which set seed most freely.
3. Learn the normal time and method of blooming of the
flowers and the length of time that the pistil will remain recep-
tive and the pollen grains capable of functioning.
4. Procure the necessary instruments, and see that these are
of an efficient kind for the work to be undertaken.
5. Be careful not to injure the flowering parts any more than
is necessary. Do not remove the surrounding flower parts, i.e.,
petals, glumes, etc., unless necessary.
6. A few crosses carefully made are of much greater value than
many pollinations carelessly executed.
60
TECHNICS IN SELFING AND CROSSING 61
Some of the common methods employed in selfing and crossing
different crops will be outlined.
Corn. In selfing and crossing individual plants of corn, vege-
table parchment and paper bags are commonly used to cover the
ears and tassels. At Minnesota, ear bags made of 40-lb. vege-
table parchment paper, 4 by 2J^ by 11 in. in size, with round
bottom, 1-in. lip, and 1-in. bottom fold, sealed with a double
strip of casein glue, have proved very satisfactory. These bags
are placed over the ears before the silks emerge and are clipped
with a collette paper clip to the stalk. After emergence of the
silks, another bag, made of extra-heavy kraft paper, 7 by 4^ by
16 in., with round bottom and 1-in. lip, is placed over the tassel.
The end of the bag is folded tightly around the stalk and held
in place with a paper clip. The next day the ear bag is removed,
and the pollen that has collected in the tassel bag is poured over
the silks of the ear to be selfed or crossed. In some cases, it is
desirable to clip off the young silks at about the time that the
tassel bag is placed over the tassel. This ensures a tuft of silks
of similar length at pollination time. After pollination, the
parchment ear bag is replaced and tied to the stalk with a string.
This bag is left on until harvest. The used tassel bag is discarded.
A method used commonly is to cover the ear shoot with a
glassine bag, approximately 6^ by 2^ in., before the silks
appear. These bags are placed over the ear shoot but are not
clipped or tied to the plant. This method works satisfactorily
when the ear shoot is large enough to support the bag. In some
early varieties and inbreds of field corn and some strains of
sweet corn and popcorn, the ear shoot is not sufficiently well
developed to hold a glassine bag in place. After the silks
appear, a specially treated tassel bag that is resistant to moisture
and weathering is placed over the tassel and held in place with a
clip. At pollination, the bottom of the glassine bag is clipped off,
the silks are pollinated, and the ear shoot, after pollination, is
covered with the tassel bag that is clipped in place.
Another method used in selfing is the " bottle method"
developed by Jenkins (1936). Small glassine bags are placed
over the ear before any silks appear. After emergence of the
silks has begun, a 2 oz. bottle of water is hung on the stalk at
the ear-bearing node with a bent wire. The tassel is cut from
the stalk, its shank is inserted in the bottle of water, and tasseJ
62 METHODS OF PLANT BREEDING
and shoot are enclosed in a large paper bag. The tassel should
FIG. 8. Details of wheat inflorescence.
Upper left, normal spikes; lower right, emasculated spike; 2, spikelet natural size; / and
g, flowerless glumes; k and r, florets; 3, a single flower closed just after flowering, 3X;4A,
longitudinal diagram before flowering, 1 x 2.5 X, a ** anthers, o = ovary, a = stigma, / =
filament; 4B diagram after flowering; 5 = transverse floral diagram, 6X, fff "* lemma,
p = palea, a = anthers, s = stigma; 6, flowerless glume, 7, lemma, 8, palea, slightly
reduced; 9, lodicule, 4X; 10, cross-section anther, 26 X: 11, pollen grains; 12, ovary and
stigma just prior to flowering; 13, at flowering; and 14, shortly after; 15, 16, 17, the mature
seed. (After Babcock and Clausen, 1918, after Hays and Boss.)
be arranged directly above the ear shoot. The bottle of water
seems to keep the tassel alive and shedding pollen as new silks
TECHNICS TN SELFING AND CROSSING 63
emerge. After 48 to 72 hr,, the tassels may be removed and
the bottles collected.
When large quantities of seed are required, as in sib pollina-
tions or crossing, it is usual to mix the pollen collected from
several plants of one line and apply the mixture to the silks of the
desired number of plants in the female line. For this purpose, a
small " pollen gun" or small insect duster may be used to apply
the pollen, the anthers first being screened out.
Large-scale production of crossed seed is accomplished by
planting the lines to be crossed in alternate blocks in a field
isolated from other corn and removing the tassels of the female
line before pollen sheds or before the silks appear. The seed
produced on the detasseled line is hybrid seed. The ratio of
pollen-parent rows to detasseled rows varies from, 1:2 to 1:4,
depending on the pollen-producing ability of the male line.
Wheat, Oats, and Barley. Hayes and Garber reviewed
studies of blooming with wheat that emphasize the importance
of this knowledge in relation to time of emasculation. The
period from about 5 p.m. to 7 a.m. was referred to as night.
Of 2,977 flowers studied on 69 spikes, 1,492 bloomed at night
and 1,485 during the day. These data show that with wheat
it is equally satisfactory to pollinate during the day as in the
very early morning.
The same writers reviewed studies that have been made to
determine whether it was necessary to cover emasculated spikes
of wheat. All results showed that emasculated spikes when left
uncovered without hand pollination set a high proportion of seed.
Crosses in the self-pollinated cereal grains may be made
either in the field or greenhouse. All but about 8 to 15 of
the florets on a spike or panicle arc removed, before anthesis,
from the heads of each plant to be used as a female parent.
The stamens are then removed from these remaining flowers
with small forceps before the anthers dehisce, and the head
is enclosed in a paper bag. Bags 2% in, wide and 6 in. long,
made of vegetable parchment paper, are satisfactory. About
2 days later, ripe anthers are collected from the male plants,
and pollen is applied to the flowers of the female by breaking
a mature anther and placing it within the emasculated floret.
The paper bag is placed over the pollinated head and allowed
to remain until harvest. Seed from both male and female
64 METHODS OF PLANT BREEDING
plants used in the crosses may be harvested also. Direct
comparisons of the progeny of the two parents with the Fi and
segregating generations of the crosses are highly desirable when
genetic studies are to be made.
Suneson (1937) found that chilling wheat plants for periods of
15 to 24 hr. at 27 to 36F. resulted in a marked reduction of self-
fertile florets through killing of the pollen. Different varieties
varied in tolerance to chilling.
Since the glumes of self-unfruitful florets are held open by the
lodicules, rapid application of the desired foreign pollen by dust-
ing on the exposed stigmas was possible. Self-fertile florets
tend to remain closed and can be rogued. This method of
emasculation might be useful if large numbers of hybrid seeds
were needed. If the female parent possesses a simple recessive
character and the male the dominant allele, any plants from
self-fertilized seed can be rogued the year the Fi plants are grown.
Rye. In self-pollinating rye, several heads of a plant may be
enclosed in a parchment bag before anthesis. It is desirable to
place an eyelet in the top of the bag and to tie it to a stake for
support. Bags of the same size as those used for ear bags with
corn are satisfactory. These bags are left on the plants until
harvest. In making controlled crosses between individual
plants, the same technic used in emasculation and pollination
with wheat, oats, or barley may be used. In making inbred-
variety crosses, the flowers of the inbred lines may be emas-
culated in the usual manner, and when the florets open 1 to 4
days later a large amount of pollen may be collected by enclosing
large numbers of heads of the open-pollinated variety in a large
paper bag and the pollen applied to the emasculated flowers
with a camePs-hair brush.
Flax. Emasculations of flax are commonly made in mid-to-
late afternoon. At this time a little experience will indicate
which flowers will open the following morning. The petals of the
flowers may be pulled out and the anthers pushed off with a
toothpick. The next morning, flowers are collected from the
male plants, and, by holding them between the thumb and fore-
finger, the dehiscing anthers may be brushed over the stigma.
It does not appear to be necessary to bag the flowers.
Cotton. In hybridizing cotton, it has been found that a short
section of ordinary soda-fountain straw, closed at the upper end,
TECHNICS IN SELFING AND CROSSING 65
may be used to enclose the exposed pistil after emasculation.
Humphrey and Tuller (1938) described an improvement in the
use of this technic. They found it unnecessary to remove all
the anthers in emasculation, since those not removed were cut off
by the straw when this was inserted over the staminal column.
The soda straw, before use, was closed at one end and about
one-fourth of the anthers frotia a male flower scooped into the
straw. This was then inserted over the pistil of the emasculated
flower, forced down until it reached the ovary, and the straw
fastened to the stem with No. 26 copper wire. By the use of
this technic the flowers were emasculated and pollinated the same
day, the flowers being worked with only once.
Sorghum. Stephens and Quinby (1933) suggested the use of
hot water for bulk emasculation of sorghum. If large amounts of
hybrid seed were needed or backcross populations desirable, the
ordinary methods of individual floret emasculation were too slow.
They found that immersion of the sorghum heads in water, the
temperature of which was 42 to 48C., for 10 min. resulted in
killing of the pollen. The equipment consisted of a large rubber
tube that could be placed over the head to be treated and tied
to the peduncle of the head at the lower end. A metal container
was connected to the upper part of the tube, into which the hot
water was poured. When proper temperature conditions were
obtained, all pollen in the head was killed. The emasculated
heads could then be pollinated with pollen of the desired male
parent.
Rice. Jodon (1938) found that immersion of the heads of
rice in water at 40 to 44C. for 10 min. destroyed the viability of
pollen without injury to other floral organs. Treatment at to
6C. gave similar but probably less effective results.
A large-mouthed Thermos jug was used as a water container
and the treatment applied in the morning prior to normal
blooming. Emasculation by hot or cold water eliminated injury
to the glumes, and the florets opened in a normal manner.
Normal seed, which germinated well, was obtained when florets
so emasculated were pollinated.
Another method used in artificial hybridization is ibo emascu-
late by removal of the anthers with small forceps through a
. slanting opening made by clipping away a portion of the upper
part of the lemma. This is done in the evening or morning
66
METHODS OF PLANT BREEDING
prior to blooming, before the anthers will shed pollen on handling.
The florets are pollinated by breaking mature anthers within the
emasculated floret.
Potato. For careful genetic experiments it is probably wise
to enclose the flower clusters in small cloth bags in selfing the
potato. Otherwise, bagging
seems unnecessary. Emascu-
lation is accomplished by re-
moving the anthers with a
small forceps or by scraping off
the anthers with a small knife.
Pollination is accomplished by
tapping a flower of the male
parent gently so that pollen is
spilled onto the thumbnail and
then applied to the stigma of
the emasculated flower.
Pumpkin and Squash. Most
varieties of pumpkins and
squashes have imperfect flowers,
some flowers on the plant having
only male and others only
female organs. Pulling the
petals of the female flowers
together so that they com-
pletely cover the stigma and
putting a rubber band around
them are easy and acceptable
ways of excluding foreign pollen.
In selfing, or crossing, the male
flowers may be collected and
pollen shaken directly onto the
stigma or first shaken onto the
thumbnail and then transferred
to the stigma.
Onion. Onions may be selfed by enclosing the head with a
paper bag before any pollen is shed. Shaking the plant daily
or tying the bagged head to a stake so that the wind will do the
shaking was found by Jones and Emsweller (1933) to increase
the amount of seed set. For seed increases, large cheesecloth
4
Fits. 9. Structure of squash flower.
1. Female flower: (a) corolla; (&)
calyx; (c) fruit.
2. Male flower.
3. Male flower with calyx and corolla
removed.
4. Female flower with calyx and
corolla removed showing: (a)
stigma; (6) style; (c) point of at-
tachment of calyx and corolla; (d)
undeveloped fruit.
5. 6. Longitudinal and cross sections
of fruit.
Size: 1, 2, % X; 3, 4, H X; 5, 6 greatly
reduced. (After Hayes and Garber.}
TECHNICS IN SELFING AND CROSSING 67
cages may be used, several plants being enclosed. The authors
mentioned above used cloth cages 3 by 3 by 6 ft. in making
crosses. Fly pupae were introduced into the cages to act as a
means of transferring pollen from one plant to another. Hybrids
of some crosses may be distinguished in the seedling stage. In
other crosses, it was necessary to grow the bulbs before roguing
out the selfcd plants. If the hybrids are not distinctly different
from the female parent, it would be necessary to emasculate the
female-parent flowers.
Red Clover. In selfing red clover, cloth bags about 4 in. long
and 2 in. wide may be used to exclude insects. Such bags are
tied on before the flowers open. Usually a higher amount of
selfed seed is obtained if the heads are rolled several times during
the flowering period to aid in tripping the flowers. Cloth bags
are preferred to paper, since the heads can be rolled without
removing the bag. These bags may be made of theatrical gauze,
such bags having a coarse mesh. Controlled cross-pollinations
may be made by hand or through the use of bees. For hand
pollination, Williams (193 Ic) states that emasculation is not
necessary, since most red-clover plants are self-incompatible.
Under Welsh conditions, the percentage of self-fertility varied
from 3.45 to 0.17 per cent, with a mean of 0.85 per cent, during an
8-year period. Partly folded triangular pieces of cardboard 2 in.
long and about }/% in. wide at the broad end and tapering to a
point at tho other are used for tripping the flowers and applying
pollen. One card is used for tripping the flowers and the other
for collecting pollen from the male parents and applying to the
stigmas of the female parents. One collection of pollen usually
will pollinate from 15 to 25 florets. After pollination, the polli-
nated heads are enclosed in cloth bags.
In using bumblebees for cross-pollination of red clover, the
plants to be crossed are grown in pots. The parents to be
crossed may be placed in insect-proof compartments in a green-
house or in the field, the former being the more satisfactory.
The bees are trapped in large test tubes (Williams 1931) and
washed for about ]^ min. by partly filling the tubes with water
and shaking, the water being changed several times. The water
is then poured off and the tubes placed in a rack for 2 or 3 min.,
after which the bees are rinsed two or three times before being
placed in a wooden box to dry.
68
METHODS OF PLANT BREEDING
1
For descriptive legend see page 69.
TECHNICS IN SELFING AND CROSSING 69
In making paired crosses, the bees are introduced into a com-
partment containing the plants in about the half-bloom stage.
After 4 to 7 days, the bees are removed, washed, and used for
other crosses. When mass crosses are made, one bee to each six
or eight plants is introduced into the compartment containing
the plants and left until the flowering period is over. Similar
methods have been found by Atwood (1940) to be satisfactory
with white clover.
Alfalfa and Sweet Clover. Several immature flowering
branches of sweet clover may be covered with a lightweight
cheesecloth bag approximately 6 or 8 in. wide and 12 in. long.
This excludes bees and prevents cross-pollination. There are
three types of plants of Melilotus alba, according to Kirk and
Stevenson: (1) those that are spontaneously self -pollinated and
self -fertile and produce seed without manipulation; (2) those that
are self -fertile but that are not normally self-pollinated without
manipulation; (3) self-sterile plants. To ensure self-pollination,
when the plant is self-fertile, it is desirable to manipulate the
bag by rubbing gently with the hands every day or two at the
time of pollination. M. officinalis is not entirely self-sterile,
and selfed seed can be obtained by the same means (Pieters and
Hollo well, 1937).
Kirk (1930) devised the suction method for emasculating
sweet-clover flowers. If the plants are located near a water
faucet, a vacuum flask inserted in the hose line will furnish the
necessary suction. Otherwise, an electric or gasoline-driven
suction pump must be used. A short piece of glass tubing
slightly less than 1 mm. in diameter is inserted in the end of the
hose. The point of this nozzle must be smooth, in order not to
injure the flower. The amount of suction is of considerable
FIG. 10. Structure of alfalfa flowers.
1. Branch showing flowers in position.
2, Single flower, showing a, standard; 6, sexual column in contact with
standard; c, keel; d, wings.
3. Seed pod.
4. Flower parts in position a, undeveloped pod; 6, ovary; c, filament;
<i, anther.
5, Same with all anthers removed except one to show stigma,
6, Anther.
Size: 1, about K X; 2, about 2 X; 3, about ^ X; 4, 5, 6, greatly enlarged.
70 METHODS OF PLANT BREEDING
importance. All the flowers on a raceme are removed except
about 20 per cent of the flowers that have most recently opened.
The petals are next removed with forceps. This ruptures' the
stamens and scatters the pollen. With the use of the nozzle
attached to the suction flask or pump, the anthers and adhering
pollen are sucked off. The stamens should be approached from
the side of the staminal tube in order not to draw the pistil and
staminal tube into the end of the nozzle. After the stamens have
been removed, the end of the nozzle should be passed over the
surface of each style and stigma, sepals and axis of the rachis.
If the operator wears a low-powered binocular magnifier on his
head, he can check on the thoroughness of the emasculation
while leaving his hands free. The pollen may be applied effec-
tively with the end of the thumbnail.
Kirk found the degree of effectiveness of emasculation by suc-
tion to be 87 per cent but suggested that improvement in technic
might increase this materially. He used the suction method on
alfalfa also with good results.
Tysdal and Garl (1938) found that when suction alone was
used and no foreign pollen applied to the stigmas, 14.1 per cent
of the flowers formed pods. If suction plus washing with a
stream of water was used, the percentage of flowers forming pods
without application of foreign pollen was reduced to 5.5.
Tysdal suggested the use of alcohol as an agent for killing
the pollen on the flowers of the female plants. The standards
were first clipped from flowers in full bloom with a sharp scissors
and the flowers tripped, leaving the stigmatic column exposed
for treatment. All flowers on a raceme to be emasculated were
treated in a similar manner. The raceme was then immersed
for 10 min. in a beaker containing 57 per cent ethyl alcohol.
The raceme was rinsed for a few seconds in another beaker
containing water, after which the adhering water was blown
off the stigma with a dentist's syringe or bulb from an atomizer
and pollinated with the desired pollen. By the use of this
method, the percentage of flowers forming pods without applica-
tion of foreign pollen was 0.89. The percentage of flowers
forming pods when foreign pollen was added was 26.3, as com-
pared with 60.0 for the suction method. Emasculation by
the use of alcohol was more complete, faster, and simpler than
emasculating by suction.
TECHNICS IN SELFING AND CROSSING 71
Grasses. Selfing, in greenhouses, may be accomplished by
enclosing a number of inflorescences in a paper bag prior to
pollen shedding. Bagging should be done soon enough so that
stray pollen that fell on the flowering panicle or spikes before
bagging will not remain viable long enough to effect cross-fer-
tilization. Glassine or vegetable-parchment-paper bags are
satisfactory. In closing the mouth of the bag, the stems are
wrapped with cotton and the bag tied over this cotton. This
excludes insects, if present, and helps to protect the stems from
injury. The upper part of the bag is tied to a stake by means of
a string inserted through an eyelet.
Bagging in the open requires careful consideration of possible
damage due to wind and rain, as well as complete exclusion of
FIG. 11. Studies of the effects of self-fertilization with grasses at the U.S.
Department of Agriculture Regional Pasture Research Laboratory, State College,
Pennsylvania.
foreign pollen. At Minnesota, vegetable-parchment-paper bags
4 by 2,^2 by 18 in., with round bottoms, sealed with casein glue,
are used for the larger grasses, such as species of Dactylis, BromuSj
Phleunij Festuca, Agropyron, and Alopecurus. A number
of inflorescences are enclosed in a single bag. The leaves
on the upper part of the stems "are removed, cotton is placed
around the stems, and the bags are tied around the cotton pad.
The bottom end of the bag is tied loosely to a stake and the upper
end tied tightly to the same stake, a string inserted through an
eyelet put in with a small eyelet machine holding the bag at the
upper end. Such bags allow for elongation of the stems and
inflorescences. The bags are left on until harvest.
At the Welsh Plant Breeding Station, vegetable parchment
that took the form of a sleeve (topless and bottomless) fitted
72 METHODS OF PLANT BREEDING
over a wire spiral was used. The wire spiral provides protection
against storms. This sleeve was fitted over the inflorescences
and tied to a stake at both top and bottom, a wad of cotton
being first wrapped around the stake and the bag tied over this.
Jenkin (1931) reported that cotton sleeves (seamless), about
15 in. in diameter and from 3 to 4 ft. long, stretched over a
frame, have been found to be highly effective in excluding foreign
pollen, provided the proper type of fabric is used. The fabric
found to be most satisfactory was a very dense and rather
heavy fabric, the most closely woven that it was possible to
procure. This fabric proved highly effective in excluding foreign
pollen but was not absolutely pollen-proof. Cheesecloth gave
very little or no protection. The cotton sleeve was tied to a
stake in a manner similar to the method of fastening vegetable-
parchment sleeves described above.
Kirk (1927) enclosed entire plants of brome grass in cotton
cages about 5J^ ft. high and 3J^ ft. square as a means of effecting
self-pollination. The bottom of the cotton cage was soaked in
oil and buried a few inches in the soil. The tops were tightly
tied. Foreign pollen probably was not absolutely excluded, but
the method was highly effective when closely woven cloth was
used. If all plants in the nursery not covered by cages were
nut off prior to pollen shedding, it would be necessary for pollen
to blow out through the cloth of one cage and in through the
cloth of another and onto the flowers before crossing could be
obtained. The amount of such cross-fertilization probably is
very small.
In making crosses by hand-hybridization, Jenkin (1924) grew
the plants to be crossed in pots and placed these in a cool green-
house some time before flowering. Emasculation was done a few
days before flowering. The upper and lower spikelets of an
inflorescence were removed and the anthers removed from the
remainder with a flat-pointed, blunt pair of forceps, the upper
florets being emasculated first. In Phleum, Alopecurus, and
PhalariSj severe thinning of the florets in an inflorescence is
necessary. After emasculation, the inflorescence is covered with
a paper bag.
Inflorescences of the male parents are covered with paper bags
prior to flowering. The greenhouse is closed tightly about an
hour before pollination begins so that any floating pollen may
TECHNICS IN SELFING AND CROSSING 73
settle down. The bag on the male plant is inclined so that when
shaken vigorously the pollen collects in the creases toward the
mouth. The bag is removed and the pollen poured on a sheet of
dark, glossy paper, previously folded into a boat shape and cut
with a sharp point at one end. The pollen is brushed lightly
over the stigmas of the emasculated flowers and thfe female unit
rebagged. Since all florets do not open on the same day and
flowering proceeds progressively downward, pollination is
repeated every day until no more f^esh stigmas are produced.
Jenkin (1931c) reported successful crosses, by the foregoing
method, with species of Lolium, Festuca, Arrhenatherum, Dactylis,
ij and Alopecurus.
CHAPTER V
THE PURE-LINE METHOD OF BREEDING NATURALLY
SELF-POLLINATED PLANTS
EARLY STUDIES
This method has been developed as a result of fundamental
studies like those of Vilmorin, Mendel, Johannsen, and of numer-
ous workers in recent times. These studies, together with field
experience, have led to the conclusion that the progeny of an
individual plant selection with self-pollinated crops may be
expected, for the most part, to breed true immediately.
A brief review of some of the more important of these early
studies will be of interest.
Le Couteur, in the early part of the nineteenth century, was a
farmer on the isle of Jersey who was interested in the problem of
improving his crops. Professor La Gaska, from the University of
Madrid, visited Le Couteur and pointed out numerous differ-
ences in plant type occurring in Le Contour's wheat field. Selec-
tions were made and the progenies tested. Some proved
superior to the commercial variety and were of more uniform
habit of growth; other selections were of little value. Bellevue
de Talevera, one of these selections, was of commercial value for
many years.
Patrick Shirreff, a Scotsman, carried on selection with wheat
and oats at about the same period as Le Couteur. He used the
individual-plant method, selecting strong, vigorous plants in his
wheat and oat fields, keeping the progeny of individual plants
separate, and increasing the more desirable. Like Le Couteur, he
proceeded orx the assumption that the selected single plants would
breed true. New varieties produced by this means were grown
extensively.
Hallett began selection with wheat, oats, and barley about
1857, believing, apparently, that acquired characters were
inherited and that improvement induced by favorable growing
conditions would be transmitted to the progeny. He raised his
plants, therefore, under the most favorable cultural conditions,
74
THE PURE-LINE METHOD OF BREEDING 75
selecting the best seed on the best developed head of the more
vigorous plants, replanting, and following the same plan of selec-
tion in subsequent years. New varieties were introduced, the
best known being Chevalier barley. Although the method
appears less desirable than that of Shirreff and there is little
reason to suppose that the continuous selection was of value in
isolating new heritable variations, it gave an opportunity to study
progenies during different seasons and in this way to select the
best. New varieties were introduced that proved of value.
The Vilmorins, in France, were early leaders in improvement of
plants by selection. Louis de Vilmorin (1856) developed the
progeny test with reference to sugar beets. Early wheat selec-
tions were made also, and the method developed is known as
Vilmorin's isolation principle. Briefly, this consists of the fact,
well known today, that the only sure means of knowing the value
of an individual-plant selection is to grow and examine its prog-
eny. Methods were developed for the determination of the sugar
content of individual roots of sugar beets. Louis de Vilmorin
observed that the progeny of some beets of high sugar content
gave progenies of high sugar content rather uniformly, whereas
others gave progeny of both high and low content and still others
gave progeny that were uniformly of low sugar content. Four
varieties of wheat were propagated for 50 years by selecting the
best plants each year. At the end of the selection period they
were compared with specimens saved at the beginning of the
experiment, and no change was noted.
Newman (1912) made an interesting review of plant breeding in
Scandinavia. The Swedish Seed Association, formed in 1886,
had a marked influence on the development of plant-breeding
methods. Hjalmar Nilsson became the director of the associa-
tion in 1891. From the beginning, careful records were kept,
individual plants were classified on the basis of minute botanical
differences, and seed of plants containing the same characteristics
was combined, the progeny of each separate type being grown in a
separate plot. Some progenies appeared so uniform that they
were especially noted by Nilsson. From a study of the records,
it was learned that in each case these were from seed of an indi-
vidual plant, there being only one representative of that mor-
phological group. This led, naturally, to the individual-plant
method of selection.
76 METHODS OF PLANT BREEDING
W. M. Hays started his plant-breeding program in Minnesota
in 1888 and from the beginning used the individual-plant method
of selection. Besides making practical studies, he initiated many
experiments that had as their purpose the formulation of funda-
mental principles. He developed the centgener plan of plant
breeding. The first step consisted of selecting individual plants
of promise, threshing these separately, and making nursery
trials of their progeny. During the period of study, plots of 100
plants each were grown from each selection. Besides taking
notes on yield and other characters on the plot basis, the 10 better
plants in each plot were selected in the field, threshed individually,
and the seed of the 5 that were of greatest promise, after labora-
tory study, was bulked and used for the following year's centgener
plot. One of the difficulties of the method as a yield trial was
that when numerous selections were made it took several days
to plant the nursery and, since only one plot of each selection was
grown, as a rule, the data obtained were not comparable. The
types of greatest promise, however, were quickly isolated and
grown in larger plots. Improved Fife, Minnesota 163, and
Haynes Bluestem, Minnesota 169, were valuable new varieties
of spring wheat selected by this method and grown widely in the
early part of the present century.
THE PURE-LINE THEORY
The experiences of plant breeders played their part in develop-
ing breeding methods, but it remained for Johannsen to place the
individual-plant method of selection on a firm scientific basis.
Johannsen (1903, 1909) made his studies with beans, selecting
this plant because it belongs to the self -pollinated group and con-
tains characters that are easy to measure. He hoped to control
heredity by applying Galton's law of regression, i.e., that the f
progeny of parents above or below the average tend to revert to
the average type. The tendency to regression toward the aver-
age could be measured and expressed statistically. By selecting
extreme parents, continual improvement could be obtained, if
the same degree of inheritance was obtained in later generations.
In studying size of beans, Johannsen found a different regression
value from that obtained by Galton and less progressive improve-
ment than he expected. This led to a study of the progeny of
individual plants, each of which varied around its mean. He
THE PURE-LINE METHOD OF BREEDING 77
found each of these progenies to be a single hereditary line,
within which there was complete regression to the mean of the
line when extreme parents were selected and their- progeny
studied. These principles are well understood today and have
had a profound effect on plant-breeding practices. Johannsen
defined a pure line as the descendants of a single, homozygous,
self-fertilized organism. Jones gave a definition, which is in
common use today, by stating that a pure line comprises the
descendants of one or more individuals of like germinal constitu-
tion that have undergone no germinal change.
THE PURE-LINE THEORY IN ITS APPLICATION
Although many experiments have been carried out that prove
that continuous selection in self-pollinated crops, as a means of
obtaining further improvement, is not worth while and there is
general appreciation of the fact that the initial individual selec-
tion is of greatest importance, there is a growing body of evidence
that heritable variations are more frequent than was supposed at
one time. An illustration may be useful. Several years ago,
Victory oats, originally produced in Sweden from an individual-
plant selection, was on the recommended list of the Minnesota
Agricultural Experiment Station. A large number of individual-
plant selections were made and their progeny studied by R. J.
Garber. When these various lines were compared for differences
of plant-breeding importance, in replicated rod-row trials, no
new line was obtained that was appreciably superior to the com-
mercial seed of Victory that had been distributed to Minnesota
farmers. The different selections showed, however, numerous
minor, heritable differences of distinct morphological type as well
as differences in quantitative characters that were more difficult
to evaluate exactly.
In this connection, recent papers by East (1935a,6,c, 1936a)
deserve consideration. He classified gene mutations under two
categories: " physiological defectives" and nondefective genes.
The former are the genes that have been used, largely, in genetic
experiments. An illustration may be given of East's viewpoint
by reference to the ligule that is characteristic of the entire
group of Gramineae. Liguleless stocks are known in maize, rye,
wheat, and oats, and in some cases the difference between liguled
78 METHODS OF PLANT BREEDING
and liguleless types behaves as if controlled by a single factor
pair. East suggests that the ligule " presumably is the result
of a very large number of non-defective mutations in various
genes, and, physiologically speaking, is the end product of a
long chain of reactions." A single mutation breaks this chain,
and a liguleless plant results.
East states that he believes every experienced plant breeder
will agree with the statement that " non-defective gene mutations
are frequent in Nature, but are difficult to detect." He sum-
marizes the results with tobacco, where characters of plants in
self-fertilized lines were evaluated statistically. Although there
was a rapid approach toward uniformity and gross homozygosis,
there still remained considerable variability, a part of which was
proved to be heritable. High mutation frequently is believed to
be responsible for heritable changes in these small genetic factors
of the nondefective type.
East gives an illustration of several cases in which, in attempts
to produce certain species hybrids, only maternals resulted. The
plants were ordinary fertile diploids and, presumably, arose from
mature gametes in which parthenogenesis was induced. The
plants then would be completely homozygous. In cases in
Nicotiana rustica, each progeny row was astonishingly alike, more
so than "any ordinary inbred populations that I had ever exam-
ined." Several of these lines were continued by self-fertilization
and within 3 or 4 years were as variable as ordinary inbred
populations.
For practical purposes, the pure-line theory furnishes a basis for
the isolation of types that differ appreciably in heritable char-
acters, and the progeny of individual plants in self-pollinated
crops breed relatively true. Mutations do occur, and minor
mutations of a nondefective type are relatively frequent, although
often not sufficiently large to be of major selection value.
Natural crosses are more frequent than is generally appreciated
and furnish another basis for variation among plants within a
variety or strain. Mechanical mixtures occur also. These
various causes emphasize the necessity of constant care to ensure
the necessary uniformity desired in an improved variety. They
do not detract from the value of the pure-line concept in its
application to the improvement of self -pollinated crops by indi-
vidual-plant selection.
THE PURE-LINE METHOD OF BREEDING 79
METHODS OF IMPROVING SELF-FERTILIZED PLAKTS
BY INDIVIDUAL-PLANT SELECTION
The following condensed summary of methods will serve as a
basis for a plan with particular crops. It is stated in general
terms, for it is recognized that such widely different plants as
tomatoes, tobacco, rice, and wheat must be grown according to
their special adaptations. With crops such as tobacco and
tomatoes, individual plants will be separately spaced in rows or
plots, whereas with the small grains bulk sowing of seed may be
practiced from the beginning of the trials.
Two principal sources of selections are available in the produc-
tion of new varieties:
1. The introduction of improved or relatively unimproved
strains and varieties of crops found in use over a wide range of
conditions, both foreign and domestic.
2. Well-adapted local varieties that are found to be variable
and to contain a composite of a number of biotypes. These may
have had their origin aajhybrids or .pure lines) which have become
altered as to general type through mechanical mixtures, natural
crossing, or mutations.
/
I. UTILIZATION OF INTRODUCTIONS
A. Source of materials.
1. A list of introduced crops and varieties with their descriptions may be
obtained from the U.S. Department of Agriculture, Bureau of Plant
Industry. It will be desirable to determine through the bureau,
when possible, the adaptation range of the introductions that appear
to meet the needs and secure seed from this source.
! 2. Personal contacts with foreign and domestic visitors is a natural
source of introductions for crops that are developed along special lines
but that frequently do not arrive through the channels of the Bureau
of Plant Industry. The visits of staff members to foreign and
domestic stations likewise may occasionally bring to attention special-
purpose crops and their varieties.
3. Mutual exchange of crops with domestic stations is a desirable
practice. Station publications afford a description of crops in use.
t 4. A survey of farm varieties is desirable. Native species^ especially
with forage crops, may yield a source of new material.
B. History and records of all introductions.
It will be desirable to keep a record book or card file recording as follows:
1, History of each introduction.
2, Description of same.
3, Year introduced.
80 METHODS OP PLANT BREEDING
<7. System of records.
A system of numbering new varieties that ensures ease of interpretation
and accuracy of record is desirable.
1. The Minnesota method is presented here and, for comparison, nota-
tions for introductions, selections, and hybrids are included.
Minnesota Records
1-20-1 Selections
11-20- 1 Crosses
III-20-1 New introductions
In this method, I, II, and III stand for individual-plant selections,
crosses, and introductions, respectively; 20 represents the year in
which the selection, cross, or introduction was made; and the final
number represents the particular selection, cross, or introduction.
Crosses are given a selection number only after having been shown to
be homozygous. Parental and F n populations are numbered by
carrying row numbers for the current and preceding season, until
homozygosity is reached. When the method of carrying row numbers
for 2 years is used in the planting plan, a pedigree can be completed
when desired. The method often used by workers in the U.S.
Department of Agriculture or in state experiment stations is given
here where FI = A, F 2 = A-l, A-2, etc., F 3 = A-l-1, A-l-2, etc.,
according to the number of selections grown.
First year = 11-18, A.
Second year = 11-18, A-l, A-2, etc.
Third year = 11-18, A-l-1, A-l-2, etc.
Selections of crosses, when given series numbers, after reaching
homozygosity, are designated 11-18- 1, II-18-2, etc., according to the
number of selections made,
2. Alberta system modification of method 1.
/ = introduction.
S selection.
H = hybrid.
Otherwise the method of numbering is similar to that used in Minne-
sota.
D. Observation of introductions.
1. All introductions may be placed under observation in small plantings.
Some of the original seed should be retained in case of unfavorable
growing conditions the first season and for later comparison in the
case of selection.
a. Plots will consist of single short rows for small grains; other types
of small plots may be used for other crops.
6. The first observations will be concerned especially with characters
of outstanding known value or for adaptability, uniformity, and
general utility of the crop. The new introductions will be com-
pared with standard varieties that are sown as frequent checks in
these observation tests.
THE PURE-LINE METHOD OF BREEDING
81
c. This preliminary planting may serve also as a means of seed
increase for larger trials.
2. During the second year, observations will again be based on small
plantings similar to those of the first year and may serve as a natural
means of eliminating those that are poorly adapted. These second
plantings are also a means of seed increase.
E. Method of testing desirable introductions for further trial (Love and
Craig 1918o, 1924) (Noll 1927) (Goulden 1931).
By means of several years of observation, a few introductions may have
been found that appear to fill a special need. The procedure of testing
these is outlined under II C.
II. PEDIGREE SELECTION WITHIN ADAPTED VARIETIES
A. /Agronomic characters sought according to needs of the regions con-
/ r
cerned.
Some important agronomic characters are given below:
1. Small grains and other cereals.
Winter hardiness
Straw strength
* Time of maturity
Drought resistance /
Quality
Nonshattering habit
2. Forage crops.
Growth habit
Quality
Drought resistance
Straw strength
Contribution to soil
content
3. Root, tuber, and sugar crops.
Sugar content
Ratio of roots to tops
Nutritive value
Seed production
Awn characters (barley)
Presence and absence of awns
Percentage of hull (barley and oats)
Seed color
'Yielding ability
Leafiness
Recovery after grazing or cutting
Cold resistance
Yielding ability (forage and seed)
organic " /Palatability
; Nutritive value
Palatability
Quality
Yielding ability
Frost resistance
NOTE: This list is not intended to be exhaustive and may be supplemented
according to the interests of the individual.
B. Resistance to diseases and insects.
1. Selection of strains resistant to diseases that are difficult to control
except through the production of resistant varieties is of greatest
importance. The following diseases may be mentioned:
Rusts Blights Root and stalk rots
Smuts Wilts Take-all
Mosaic Scab Anthracnose
2. Selection of pathogens.
a. Study the disease reaction in special disease nurseries and in the
greenhouse with the use. of physiologic races existing in the
particular locality or over a wider region.
82 METHODS OF PLANT BREEDING
b. Grow special disease nurseries in several places in the area in order
to test for resistance under field conditions to physiologic races
as they occur naturally.
3. Disease garden.
a. Selections should be tested in short rows or in other types of plots,
6. Disease epidemics.
(1) Artificial epidemics should be induced on susceptible border
rows grown throughout the nursery and generously distributed
through the plots of tested varieties or directly on the varieties
themselves.
(2) Natural or artificial epidemics may be obtained by growing the
particular crops on soil infected by wilt, root rot, etc.
4. Insect pests.
a. Grow short-row plots on soil or in regions infested with such
insects as Hessian fly, jointworms, boll weevils, borers, etc.
5. Replication frequently is important in testing for resistance to plant
diseases or insect pests. It is helpful in many cases to plant at
different periods in order to obtain favorable conditions for producing
the epidemic.
C. Technic for selection and testing.
1. Single-plant basis for selecting lines.
a. First year. Select approximately 1000 heads from individual
plants of the type desired. The total number of initial selections
depends on the crop and the amount of land and funds available
for subsequent testing.
b. Second year. Sow 25 to 50 seeds of each selected plant in space- or
bulk-planted single-plant or head-progeny rows. Discard all
plant rows that appear of undesirable type, Heterozygous types
of extreme promise may be rcselected. Continue elimination
of undesirable types in each successive year. Bulk seed of the
individuals for test in progeny rows.
c. Third year. Replication should be started this year for prelimi-
nary yield trials. Observe lines for uniformity for such agronomic
characters as date of heading, strength of straw, and height of
plant. Disease tests may be carried out as described under II B 3.
d. Fourth to sixth years. The number of years indicated is arbi-
trarily suggested, and these trials should be continued to the extent
found desirable.
(1) Composite seed of the replications of previous year's test and
grow in single- or three-row plots or in other types of plots
when desired. Replication is necessary.
(2) Plant a duplicate test in the disease garden each year, and test
for resistance to important diseases and insect pests.
(3) Test for special characters grow replications in a particular
environment, as on peat, sand, etc.
(4) Make quality tests on the crop from border rows.
(5) Select the more desirable lines for more extensive trials.
THE PURE-LINE METHOD OF BREEDING 83
e. Seventh to ninth year. Test promising lines in advanced trials,
i.e.j in J^o-acre plots replicated or in row plots with more repli-
cations than in earlier years and, when possible, at a number of
stations. In these yield trials, replicate to the extent found
necessary.
III. COOPERATIVE TESTS AND DISTRIBUTION OF PROMISING LINES
A. Bring information of the proved lines and introductions before the
farmers through the extension service, agricultural high-school teachers,
county agents, and crop-improvement associations and by means of
bulletins.
B. Select reliable farmers to grow demonstration plots of the improved lines
in comparison with standard varieties.
1. Plots, a single drill width, in the center of the farmer's field are used in
Minnesota.
2. A replicated trial may be made with a few farmers or local schools
when desirable.
C. Arrange these demonstration projects in a number of counties or prov-
inces, and organize a field day for the community at which the county
agent can use these plots as part of his program.
D. Distribute seed to those interested through the farmers' crop-improve-
ment association.
ILLUSTRATIONS OF VALUABLE VARIETIES OF SELF-POLLINATED
PLANTS PRODUCED BY APPLICATION
OF THE PURE-LINE THEORY
Selection has played a large part in the production of new
varieties of wheat, oats, barley, flax, and other self -fertilized crop
plants. Clark (1936) has given the origin of many of the varie-
ties of spring and winter wheat. In winter wheat, lobred,
loturk, and lowin, selected by L. C. Burnett, at Ames, Iowa,
have been grown extensively. Nittany, selected from the Ful-
caster variety by Noll, is the principal variety grown in Pennsyl-
vania. Nebraska 60, selected by Kiesselbach from the Turkey
variety, is grown widely in Nebraska, and Kanred, selected
from Crimean by Roberts, with an estimated acreage of 3J^
million acres in 1929, has been of great value in the hard red
winter-wheat region.
In spring wheat, the early selections, Improved Fife (Minn.
163) and Haynes Bluestem (Minn. 169) introduced about 1900
were important varieties in the early part of the present century.
Mindurn durum, the standard for quality of semolina products
and the most widely grown durum variety in United States and
84 METHODS OF PLANT BREEDING
Canada, was produced by plant selection at the Minnesota
station.
In a discussion of superior germ plasm in oats, Stanton (1936)
described many new varieties that have been developed by plant
selection. Fulghum oats and its many strains originated from a
single plant selected from Red Rustproof by J. A. Fulghum. The
single plant was earlier and taller than the Red Rustproof
variety. Other important selections from Red Rustproof and
Fulghum include Kanota, Franklin, Columbia, Nortex, and
Frazier.
Varieties of Kherson and Sixty-day oats are grown extensively
in regions of the corn belt where early oats of the Avvna saliva
group seem desirable. Gopher, a white-seeded strain of sixty-
day, has been grown extensively in southern Minnesota and in
other states where early oats are adapted. It is perhaps the
stiffest strawed early variety available. Its production empha-
sizes the ease of improvement in some cases. Only 200 original
plant selections were made from an early variety with mixed seed
color. The first year in plant rows, six strains excelled in
strength of straw, and the remainder were immediately discarded.
Gopher was the best yielding strain of the six. Richland and
logold selected by Burnett are both resistant to black-stem rust.
Nebraska 21, selected in Nebraska, has been grown widely.
State Pride, a plant selection made in Wisconsin, has been the
standard early variety in that state.
Among midseason varieties, Colorado 37 is outstanding in
strength of straw and suitability for growing under irrigation.
Cornellian, Ithacan, Upright, and Lenroc, selected by Love, in
New York, occupy about 50 per cent of the oat acreage in that
state. Rainbow and Rusota are important varieties selected
from Green Russian at the North Dakota station. Both are
resistant to black-stem rust.
Improved varieties of barley resulting from plant selection have
been given by Harlan and Martini (1936). A few of the more
widely grown varieties will be mentioned. Atlas selected from
the coast variety is the most important variety in California. In
the Manchuria-Oderbrucker group, Manchuria, Minn. 184, was
selected at the Minnesota station. Wisconsin Pedigrees 5 and 6
selected from Oderbrucker are the chief strains selected from
this variety. Peatland, selected at Minnesota, in cooperation
THE PURE-LINE METHOD OF BREEDING 85
with Harlan of the U.S. Department of Agriculture, is especially
well adapted to peat soils and is valuable also because of its
resistance to scab and black stem rust. Trebi, selected by
Harlan, was grown on an estimated acreage of 2,224,000 in 1935,
the largest acreage devoted to any single variety. It is not a
desirable malting variety, but in spite of several undesirable
characters it has high yielding ability and is especially well
adapted for growing under irrigation.
Practically all the varieties of rice grown in the United States,
according to Jones (1936), were developed by selection, although
not all were obtained by pure-line selection. More recently,
hybridization has been used as a method of breeding, but to
date only one variety produced by hybridization is grown
commercially.
Dillinan (1936) states that all varieties of seed flax grown in
the United States were obtained by plant selection. Bison,
selected by Bolley, at the North Dakota station, from commercial
seed obtained from Belgium, is the most widely grown variety in
United States. Buda, selected also by Bolley, has been a popular
variety. Redwing, selected in Minnesota, is an early-maturing
variety well adapted to southern Minnesota and Iowa, where it is
extensively grown. All three varieties are resistant to wilt.
Without this resistance it would have been impossible to continue
to grow flax in the hard red spring-wheat belt.
Individual-plant selection has been of great importance also in
peas and beans (Wade 1937). Strains of Alaska peas and of other
varieties have been selected that are resistant to fusarium wilt,
M.A.C. Robust, selected by Spragg, in Michigan, is resistant to
mosaic and has been grown extensively in Michigan and New
York. Among the present varieties of soybeans, Morse and
Cartter (1937) state that a considerable proportion were obtained
by selection from the large number of introductions that were
obtained from the Orient. Individual-plant selection has played
a large part also in the origin of tobacco varieties, as has been
pointed out by Garner, Allard, and Clayton (1936).
CHAPTER VI
HYBRIDIZATION AS A METHOD OF IMPROVING
SELF-FERTILIZED PLANTS
SOME STUDIES BEFORE 1900
Many studies were made during the eighteenth and nineteenth
century for the purpose of learning the laws of inheritance in
hybrids or to develop new and improved varieties. A few of
the more importar^t of these early studies will be mentioned to
suggest the extent of the many investigations made before the
present century, each of which played a part in developing
principles that have led to the present view of a planned plant
breeding program.
Kolreuter, in 1760-1766, made extensive studies of artificial
hybrids and emphasized especially hybrid vigor in F\ crosses.
He noted the intermediate condition of the F\ in crosses in tobacco
and interpreted this as showing the effect of the male parent.
Thomas Andrew Knight, born in England in 1759, contributed
greatly to early plant breeding. Much of his work was with
fruit crops apples, pears, peaches, currants, and grapes. He
emphasized the value of crosses as a means of obtaining new
combinations of characters. John Goss, about 1820, studied
segregation in crosses with peas but did not give an adequate
explanation of the nature of the segregation. Sargaret, about the
same period, made crosses between muskmelon and cantaloupe
and studied fruit characters in F\. He reported the appearance
of differences in flesh colors, seed color, rough or smooth fruit,
extent of ribbing and flavor and emphasized the dominance of one
character over the other. Gartner, in 1849, studied thousands of
crosses, observing the uniformity and appearance of the Fi
generation. Naudin, in 1865, just prior to Mendel's report, noted
the uniformity of the F\ generation and segregation in F 2 , ascrib-
ing this to the segregation of heritable factors in the formation of
male and female reproductive gametes.
Mendel's work need be mentioned only briefly. He studied
individual characters and placed his results on a definite factor
86
HYBRIDIZATION AS A METHOD OF BREEDING 87
basis. The methods used by him are not widely different from
those used today. Although the laws of inheritance are much
more complex than those presented by Mendel and although most
normal characters are dependent upon the interaction of many
genetic factors, the methods of work introduced by him have
found very wide application. These methods have made it
possible to develop a planned plant-breeding program based on
the laws of heredity.
William Farrer, of Australia, during the latter part of the
nineteenth century, approached present-day plant-breeding
methods and developed many wheat varieties of great value. He
selected parents for crosses on the basis of their characters,
strongly featuring the value of composite crosses as a means of
inducing maximum variation. Federation, a variety of wheat
that was early maturing, nonshattering, with stiff, erect, short
straw, was produced as the result of a definite attempt to obtain
a variety of wheat suited to gathering with a stripper.
An illustration of the crosses used in the parentage of Federa-
tion is given below:
Improved Fife X Etitwah
I I
Yancii
ilia X Purple Straw
Federation
By similar means, he obtained the following varieties: Come-
back, Ceder, Firbank, Bobs, Cleveland, and Florence, the latter
being a bunt-resistant variety.
The work of A. P. and C. P. Saunders in Canada is well known.
In 1892, A. P. Saunders crossed Hard Red Calcutta with Red
Fife. C. P. Saunders took over the experimental work at Ottawa
in 1903 and continued the selections that led to Marquis and
other varieties, the new variety Marquis being first grown in pure
form in 1904, 12 years after the original cross was made. He
used the individual-plant method of selection and determined the
gluten quality in the progeny of crossbred wheats by the chewing
test.
DEVELOPMENT OF METHODS SINCE 1900
The rediscovery of MendePs l#ws by De Vries, Correns, and
von Tschermak, in 1900, stimulated the extensive studies of the
88 METHODS OF PLANT BREEDING
laws of heredity that have led to the present system of breeding
crop plants with a definite plan to obtain the combination of
characters desired. The first step in such a program is a careful
study of material available and an analysis of the characters
desired. The necessity of making a collection of all available
strains and varieties and an analysis of their characters needs
emphasis. Although there is general appreciation of the desir-
ability of this step, it is seldom practiced to the extent that would
seem to be worth while. In recent years, Vavilov and his
coworkers in Russia have made extensive world-wide collections
of many crops. The U.S. Department of Agriculture maintains
very extensive collections of varieties of grains, fruits, and
vegetables that serve as a potential source of material for breed-
ing. New strains and varieties are being added constantly
through the plant-introduction service. The second step is to
obtain the most desirable strains by selection. When these steps
have been taken and the necessary background of knowledge
with respect to disease reaction and agronomic characters of the
crop has been gained, a crossing program may be undertaken.
Crosses are made with a definite purpose in mind, i.e., with the
intention of combining in one variety the characters desired.
Illustrations will be given from some of the present problems
being studied at Minnesota.
BREEDING IMPROVED VARIETIES OF BARLEY
This work, cooperative between plant geneticists, plant path-
ologists, and cereal technologists, illustrates the value of a
cooperative program.
The first crosses for the Minnesota experiments, designed to
produce satisfactory smooth-awned barleys, were made in 1912.
These involved crosses of Lion, a six-rowed, black-glumed,
smooth-awned variety with good, adapted, six-rowed, white-
glumed, rough-awned sorts. Strains with white glumes and
smooth awns were selected and appeared to yield well in rod-row
tests (Harlan and Hayes 1919). In more extensive tests (Hayes
1926), it was found that in some seasons these strains were
reduced in yield because of the "spot-blotch" disease caused by
Helminthosporium sativum. This led to a cooperative attack on
the problem by plant geneticists and plant pathologists. It was
HYBRIDIZATION AS A METHOD OF BREEDING 89
found that the greater susceptibility of the hybrids appeared to be
due solely to their greater susceptibility to H. sativum.
A second series of crosses was made, with the use of one of the
better white-grained, smooth-awned segregates of the first cross
with desirable white-grained strains of the Manchuria type that
were resistant to spot blotch.
Studies of disease resistance were made in specially prepared
disease gardens. The mode of inheritance of reaction to Hel-
minthospormin was studied in the early segregating generations
(Hayes el al. 1923 and Griff ee 1925), and it was found that at
least three genetic factors were involved in differentiating reaction
to the spot-blotch disease.
Selections of smooth-awned segregates of desirable plant and
kernel type were made in the plant rows found to be resistant to
H. sativum. From the cross of a smooth-awned segregate of the
first cross with Luth was produced the variety Velvet. The
cross of " smooth awn 7 ' X Manchuria led to the production of
Glabron. Velvet is grown rather extensively at the present time.
The most extensively grown smooth-awned variety of the Man-
churia type is Barbless (Wisconsin 38) produced from a cross of
lion X Oderbrucker by Leith of the Wisconsin Agricultural
Experiment Station.
The smooth-awned varieties Velvet, Glabron, and Barbless
are susceptible to stem rust and to blight. The variety Peat-
land, selected at Minnesota from a lot of seed obtained from
Switzerland, is resistant to both diseases. Studies by Powers
and Hincs (1933) and by Reid (1938) showed that the stem-rust
resistance of Peatland was due to a single dominant factor, in
crosses with susceptible varieties. Brookins (1940) has shown
that the factor pair conditioning resistance and susceptibility to
a collection of races of rust in the field in the mature-plant stage
also controls the same type of reaction to races 19, 36, and 56 in
the seedling stage in the greenhouse.
Peatland was crossed with Barbless for the purpose of com-
bining the good characters of both. The contrasted characters
are given in the table shown on page 90.
A large F 2 was grown in a space-planted nursery and only plants
with smooth awns that were resistant to stem rust and that were
of desirable plant and seed type were selected. Since smooth
awns are due to one main recessive factor, the selection of smooth-
90
METHODS OF PLANT BREEDING
awned plants in F% eliminated all rough-awned sorts from subse-
quent generations. Resistance to stem rust is dominant.
Consequently, the ^2 resistant plants gave rise to homozygous
and heterozygous F 3 progenies in a ratio of 1:2.
Character
Barbless
Peatland
Yield
Good*
Fair
Seed size
Large *
Moderately small
Type of awns
Smooth*
Rough
Strength of straw
Poor
Fair*
Resistance to stem rust
Resistance to scab and
blight
Susceptible
Susceptible
Resistant *
Resistant *
Resistance to loose smut . .
Resistance to covered smut
Resistance to stripe
Susceptible
Moderately resistant*
Moderately resistant*
Resistant *
Resistant*
Susceptible
Resistance to spot blotch
Moderately resistant*
Moderately resistant*
* Indicates desired character.
Tests of reaction to blight were begun in F 3 or F in a special
tent under conditions conducive to the development of a severe
epidemic. Notes on reaction to blight are taken just before the
heads ripen. These notes are then used in conjunction with
notes on reaction to stem rust and smut, date of heading, height
of plants, and lodging in conjunction with observation of general
vigor and appearance of the plants in making individual-plant
selections. Many of these strains are now in rod-row trials and
appear promising.
Other crosses are, of course, made also and carried along in a
parallel manner. One of these involves Velvet X Chevron.
Chevron is a sister selection of Peatland and has a similar reaction
to the common barley diseases. If some of the best segregates
are found to be a distinct improvement over the existing varieties,
they will be released for distribution to the farmers. The best of
these, from different crosses, may then be crossed in an effort to
bring about further improvement.
Tests of yielding ability begin with rod-row trials at University
Farm only. The best of these are then tested in replicated yield
trials in four stations in Minnesota. After a 3-year test in rod
rows, the most promising strains are then tested in J-^Q -acre-plot
HYBRIDIZATION AS A METHOD OF BREEDING 91
trials in six places in Minnesota, usually for a period of 3 years,
before a final conclusion regarding distribution to the growers is
made.
Studies of diastatic activity are made on the material in rod
rows in cooperation with the Division of Agricultural Biochemistry
at the University of Minnesota. When the new strains go into
J<40-acre-plot trials, a complete malting test is made in the cooper-
ative malting laboratory at the University of Wisconsin.
BREEDING BY HYBRIDIZATION
Examples have been given above and in previous chapters of a
few problems that are being attacked or that have been solved
through the crossing of different varieties and the combination of
characters from them. The broad principles involved in a
hybridization program, the selection of the parental material,
and a general description of the method of handling the hybrid
material will be discussed here. The detailed outline of the steps
to be followed in successive generations following the cross will be
given under Methods of Breeding.
Object of Crossing. The object of crossing is to combine in a
single variety the desirable characters of two or more lines,
varieties, or species. Occasionally the recombination of genetic
factors leads to the production of new and desirable characters
riot found in either parent. In a planned program every effort
should be made, however, to select parents that have the char-
acters desired. Frequently transgressive segregation occurs for
quantitative characters such as yield, height of plant, earliness,
and resistance to lodging. Selection of parents that are already
relatively satisfactory for these characters will enhance the
probability of obtaining the desired end result.
Selection of Parental Material. The procedure to be followed
in selecting parental materkl for crosses will depend upon the
extent to which the station conducting the breeding program has
previously experimented with the crop in question. A station
that has conducted extensive variety tests for any given crop will
in all probability have sufficient data to inaugurate a breeding
program without further study of the parental material. How-
ever, a station beginning a breeding program with a crop of which
it knows relatively little should conduct a thorough study of all
present varieties (and of species in some cases) of that crop before
92
METHODS OF PLANT BREEDING
FIG. 12. Uton oats was bred in Utah by Tingey, Woodward, and Stanton
(1941). It combines the large, white kernel of its Swedish Select parent with
resistance to smuts from its Marktoii parent.
The upper photograph shows reaction to loose smut and the lower photograph
to covered smut. The two bundles at the left of the photographs show the
proportion of smutted and smut-free panicles of Swedish Select, the two at the
right the smutted and smut-free panicles of Utou.
HYBRIDIZATION AS A METHOD OF BREEDING 93
beginning a breeding program. The importance of having a
thorough knowledge of the parental material cannot be too
strongly emphasized.
Technique of Crossing. Crossing may be performed in either
greenhouse or field. The greenhouse offers better protection
from the elements and from stray pollen and often provides more
satisfactory temperature and humidity. Crosses in the green-
house can be made at almost any time of the year.
It is advisable to make several dates of planting, particularly
when the parents differ in time of flowering. The parents should
be sown in short rows (or in pots), with the seeds individually
spaced and with sufficient space between the rows so that the
plants can be worked with easily.
A study of the structure of the flower will reveal the best
method of making the crosses. A study of the viability of the
pollen and the time of receptivity of the stigma will be of material
aid. An examination of the stigma and anthers sometimes will
reveal which of the two parents should be used as the female.
For example, it is known that it is more difficult to obtain crossed
seed when barley varieties with unbranched stigmas are used
as females than when varieties with branched stigmas are used as
females. It is important also to prevent as much injury to the
flowers as possible. If possible, use as the female the parent with
a recessive character so that selfs can be discarded when the F\ is
grown. The details of emasculation and pollination were dis-
cussed in Chap. IV.
Handling the Hybrid Material. Sufficient FI plants are neces-
sary to give the amount of seed required for the F 2. If grown in
the field, the Fi seeds should be individually space-planted far
enough apart to give maximum seed production. In the green-
house the Fi seeds are planted in pots or in the soil of the green-
house bench. Some artificial light may be necessary in the winter
time in the northern climate, and frequently a complete fertilizer
is needed.
If the pedigree method is used, the F 2 and succeeding genera-
tions, until bulked, should be grown in spaced planted progeny
rows with 25 to 50 seeds per row. In certain studies, replication
is desirable. The parental checks should be sown every 10 to 30
rows so that frequent comparisons can be made with them in
selecting plants for further study.
94 METHODS OF PLANT BREEDING
Selections for disease resistance, height of plants, date heading,
and any special characters such as head type, color of glumes, and
type of awn (in the case of cereals) are made from plant rows in
jP 3 and later generations. In selecting for leaf rust, the plants
must be marked several weeks before harvest. Selection for
stem-rust resistance can be made at harvest time. The general
vigor and habit of growth of the lines is observed in making selec-
tion and this observation used in conjunction with notes on
specific characters. The individual plants harvested are threshed
separately and the seed examined in the laboratory for type, size,
shape, color, and plumpness. Seed of plants with inferior grain
quality is discarded.
Disease tests usually arc begun in the P\ by subjecting the
plants to an epidemic and selecting only resistant plants. In F 3
and later generations, the usual method is to plant separate
nurseries in order to study reaction to the various diseases for
which it is hoped to obtain resistant varieties. Resistant plants
are then selected in these disease nurseries. In breeding for
resistance to diseases in which the methods used in inducing the
disease epidemic results in abnormal plant growth, the common
practice is to grow special disease nurseries but to select plants or
lines from a duplicate nursery grown under normal conditions,
lines found to be susceptible in the disease nursery being
discarded.
Quality tests are made whenever possible. In some breeding
programs these can be begun in F z . Usually these tests are made
in later generations when a greater bulk of seed or plant material
is available. This is true particularly when the cost of making
the quality tests is great. The material is then purified, and all
otherwise undesirable lines are discarded before being tested for
quality.
Special characters may be studied by growing the F* and subse-
quent generations under environmental conditions that will bring
out the differentiation desired.
During the segregating generations, it is desirable to grow rows
of the parents and of the best available standard variety at fre-
quent intervals throughout the nursery. Only plants and prog-
eny rows that appear equal to the standards in all respects should
be selected.
HYBRIDIZATION AS A METHOD OF BREEDING 95
^ ' ^'Wiw . , , , ,
METHODS OF BREEDING
Several methods of breeding self-pollinated crops through the
use of hybridization have proved satisfactory for particular
problems. These may be classified as:
1. The pedigree method.
2. The bulk method.
3. The backcross method,
4. Multiple crosses.
In these and other problems of a similar nature, the larger the
populations during the segregating generations the more chances
there are of obtaining the combination of characters that are
desired. The more complex the inheritance the greater the need
for larger numbers. Although the exact combination of char-
acters desired may not be recovered in F%, there is still the possi-
bility of obtaining it in F& or later segregating generations.
When two factors are closely linked, their recombination will be
obtained infrequently in F% but secured more easily in jF 3 by
growing the progeny of F% plants that contain one of the two
desired characters. In most crosses it is a sensible plan to grow
as large an F 2 population as can be sampled adequately
in'F 8 .
Pedigree Method. This method consists of (1) making a cross
between two parents possessing the characters that it is desired to
combine in a new variety, (2) growing the material in spaced
plant rows so that individual plants may be studied, and (3)
keeping a system of records so as to be able- to trace individuals
from one generation to the next. A ! Numerous systems of keeping
progeny records are available and will be chosen to suit the needs
of the investigator. An outline of some of the methods was given
in Chap. V.
The number of seeds sown, length of row or size of plot, and
number of generations grown before bulking the plants in progeny
rows will vary considerably with the different crops. For cereal
crops the following procedure will illustrate the steps involved.
1. Grow sufficient F\ plants to produce the desired amount of
seed for .PV Compare the Fi plants with the parent varieties,
note dominance of characters, and discard selfs. Seed from the
identical parent plants used in producing each Fi progeny may be
96 METHODS OF PLANT BREEDING
grown beside the F\ and, in critical studies, seed from these
parent plants grown for comparison with F% and later generations.
2. Grow 2000 to 10,000 individually spaced F 2 plants. In
F 3 and subsequent generations, grow 1000 or more progeny rows
each year from seed of individual plants selected the previous
year. Select on the row basis first, and then select the best plants
in these rows. Discard any lines found to be undesirable in
disease nurseries.
3. Bulk the seed of rows when homozygous. This usually is
done in F to F Q . At Minnesota, with small grains, pedigree
selection is continued until F b , when promising apparently homo-
zygous lines are bulked for the yield trials. Some lines may be
continued in F$ from selected F& plants before bulking. Lines not
homozygous in F 6 are discarded unless very promising. Appar-
ent homozygosity is determined by examination of the individual
plants of a line, in the field, for observable agronomic or disease
characters, and then the plants are harvested and threshed
individually and the seed examined before bulking the seed for
yield trials.
4. Conduct yield trials, and release for distribution to the
growers as described in Chap. V.
Bulk Method. This method consists of growing the material
in a bulk plot, usually from the jF 2 to about the FQ generations,
inclusive, followed by head selection in F 6 . By the F 6 genera-
tions, a high proportion of the plants will be homozygous for most
observable characters. The bulk plots can be subjected to
disease epidemics and special conditions as an aid in selection.
Natural selection probably will eliminate some of the weaker
types. The progenies of plants selected in F 6 are tested in the
manner described for improvement by selection in Chap. V.
Because of the ease with which crosses can be carried in bulk, a
greater number can be grown in this way during the segregating
generations than by the pedigree method. However, in the
absence of selection through FQ a higher proportion of the popula-
tion will be undesirable than with the pedigree method, in which
case careful selection over a period of years would have eliminated
more of the undesirable types. As a consequence, it would be
necessary to select more plants in F$ for testing in plant rows than
would be necessary to test in F& or F 6 with the pedigree
method.
HYBRIDIZATION AS A METHOD OF BREEDING 9<
Harrington (1937) suggested a modification of the bulk method
called the mass-pedigree method. This involves a combination of
the two methods. The material is grown in bulk until a favorable
season provides conditions for efficient selection. Then head
selections are made and grown the following year in progeny rows
as described for the pedigree method. The essential feature
of this method is the growing of the crosses in bulk until a year
favorable for efficient selection occurs, when single-plant selec-
tions are made and the pedigree method used from that time
onward. In order to make selections in bulk plots, selection
would need to be made on the head rather than plant basis.
Frequently a great deal can be learned regarding the genetics of
the material during the segregating generations when the pedigree
method is followed. Such is impossible with the bulk method.
Backcross Method. This method is used primarily when it is
desirable to transfer one or two simply inherited characters of the
nonrecurrent parent to the recurrent parent, which is usually a
highly improved variety of a desirable agronomic type.
An outline is given of the steps to be followed:
1. Grow the FI and backcross to the recurrent parent.
2. Grow 50 to 200 individual backcrossed plants in spaced progeny rows.
3. Select desirable individuals, i.e., those containing the characters to be
selected from the nonrecurrent parent.
4. Backcross these selected plants to the recurrent parent. Continue
backcrossing and selecting from 2 to 6 generations. In some cases, it
may be necessary to study the progenies of selected plants before making
the next backcross.
5. After backcrossing is finished, the material is handled in the same man-
ner as outlined for the segregating generations by the pedigree method.
There is, however, this difference. After several generations of back-
crossing, many of the factors from the recurrent parent will be homo-
zygous, and fewer generations need be grown in individual plant rows
before bulking. Only a few years of yield tests will be required also,
since most of the lines will be similar to the recurrent parent in all but
the one or two of the characters added from the nonrecurrent parent.
Multiple Crosses. Harlan and Martini (1940) have suggested
the use of compound crosses. The method may be illustrated by
assuming that eight varieties are to be combined. A series of
bridging crosses is made as follows: a X 6, c X d, e X /, g X h.
In a second mating, the FI plants will be crossed to produce the
double crosses (a X b) X (c X d) and (e X /) X (g X h). In
98 METHODS OF PLANT BREEDING
the third mating the double crosses will be combined as follows:
[(a X 6) X (c X d)] X [(e X /) X (g X h)]. As segregation will
have taken place at the time the second cross is made, a greater
number of crosses would need to be made than in the first mating.
In the third cross, a very large number of seeds would be desired,
since every seed contains essentially a different genotype and will
result, presumably, in a new combination of characters. This
procedure offers some promise of obtaining unusual combinations
of factors, leading to the production of exceptional segregates.
Its disadvantages would lie in the fact that several of the parent
varieties probably would be undesirable for certain characters,
and crosses between them would lead to the production of a
higher proportion of plants with these undesirable traits. Large
populations would need to be grown during the segregating
generations following the compound crosses. These can be
carried by either the pedigree or bulk method of breeding.
COMBINING ABILITY
Plant breeders observe very frequently that more desirable
segregates are obtained from some crosses than from others.
Some varieties are good parents, as judged by their ability to
transmit high yield and quality to their progeny in crosses; others
are less desirable. In the production of hybrid corn, the fact
that some inbred lines transmit higher yielding ability to their
Fi crosses than do others, when crossed with a series of inbreds,
has been known for many years. The classification of varieties of
self -pollinated crops as to whether they will transmit high yield in
crosses has hardly begun.
Harrington (1932) suggested that an analysis of the characters
that could be studied in an F% population will provide a means of
predicting the value of a cross. Harlan and Martini (1940)
crossed 28 varieties of barley in all possible combinations of two
each, making 378 crosses. These crosses were each carried in
bulk plots, without selection, until the eighth generation and then
space-planted. Plant selections were then made from each cross
and tested in progeny rows the next year. Since each variety
was crossed with each of the other 27, the potential value of each
variety, in crosses, could be determined from the average yield
of the selections made in crosses involving each parent in turn.
The varieties Atlas, Club Mariot, Minia, Trebi, and Sandrel
HYBRIDIZATION AS A METHOD OF BREEDING
produced an unusually high percentage of superior selections.
Crosses involving Glabron produced very few. Some varieties
that had not been sufficiently promising in nursery tests to be
grown in plots were found to be superior parents.
Harrington (1940) and Immer (1941) studied the yield of bulk
crosses in wheat and barley, respectively, in early generations as
a means of determining the comparative breeding value of differ
ent crosses. The study by Immer will be reviewed briefly.
Six barley crosses were compared with one another in JPi, F%,
F 3 , and jP 4 and with the parents. The yields of FI and parental
checks were determined from rows of 11 plants per cross or
variety, spaced 5 in. apart, and replicated six times. The tests
in F%, FZ, and P\ were made in five replicated rod-row plots, the
parents being included. The seed for F 3 and F* tests was a
random sample from F 2 and F 3 , respectively.
In Table 4 is given the average yield of each pair of parents for
3 years, expressed in percentage of the mean yield of parents for
all crosses. The yields in FI to Ft are expressed in percentage
of the average mean yield of the parents in the six crosses for the
year or years in which the test was made.
TABLE 4. YIELD OF PARENT VARIETIES AND F\, F 2 , F s , F '4 CROSSES IN
BARLEY, EXPRESSED IN PERCENTAGE OF THE AVERAGE YIELD
OF THE PARTCNTS GROWN THE SAME YEAR AS THE CROSSES
Cross
1938
1939-1940
1940
Average
of
parents
Fi
Parent varieties
F
F s
F 4
9
cf
Average
Barbless X Chevron
Barbloss X Minsturdi
121
118
117
87
80
76
151
186
142
116
81
89
127
127
111
127
127
111
93
74
93
109
51
51
110
101
102
118
89
81
119
125
140
137
115
111
114
114
118
124
105
101
113
120
83
111
117
100
99
Velvet X Chevron
Barbless X Olli
Barbless X C.L 2492
Velvet X C.I. 2492
Average
100
128
100
125
105
The crosses of Barbless X Olli and Velvet X Chevron pro-
duced the highest yields in ^2 and*F 3 and were among the highest
in F 4 , They were intermediate in yield in FI. The two crosses
100 METHODS OF PLANT BREEDING
involving C.I. 2492 were relatively low in yield in all four genera-
tions tested.
It appears that tests of bulk crosses in F% or F 3 may be used as a
means of discarding entire crosses in the early segregating genera-
tions, and the plant breeder then can make selections only from
the crosses that promise the greatest proportion of high-yielding
segregates. Testing in several different localities and for more
than 1 year would be advisable.
From the study by Immer, it appeared that the F\ could not be
used to determine satisfactorily the potential value of a group of
crosses. Since the amount of seed in FI is very much limited,
space planting must be resorted to. It was found that some of
the varieties and crosses responded in a differential manner when
space-planted 5 in. apart as compared with seeding in rod rows
at the regular rate for such trials.
As information on the sources of good germ plasm in crop
plants accumulates, as measured by combining ability in crosses,
the outstanding parent varieties will be isolated and used more
extensively in breeding programs.
In using the pedigree or bulk methods of breeding, the first
yield test is obtained usually in F 6 to F 8 . It would be highly
desirable to know the yielding capacity of different strains
from a cross and to discard the low-yielding ones before F G
to FV Although no experimental data are available, it would
appear that replicated yield trials in F^ from bulked seed of F 3
lines that themselves were each the progeny of an individual F 2
plant, should supply information on the relative yielding capacity
of such strains. A space-planted nursery of the same strains
could be grown the same year and single-plant selections made.
In F 5 plant-progeny rows would be grown from the strains found
to produce high yields in the yield trials. The plant-progeny
rows could be bulked in ^5 or F$, if homozygous, for regular yield
trials of pure material. By this procedure, a higher proportion
of the strains in the regular yield trials would be expected to give
satisfactory yields.
CHAPTER VII
THE BACKCROSS METHOD OF PLANT BREEDING
Baekerossing, when possible, is the most satisfactory method to
use in genetic studies to determine linkage relations. It is also
useful as an aid in developing a factorial hypothesis. Harlan
and Pope (1922) pointed out its probable value in small-grain
breeding and stated that it has been " largely if not entirely
neglected in any definite breeding programs to produce progeny
of specific types/' They suggested the probability that there
were many instances in which backcrossing would be of greater
value than the more common method of selecting during the
segregating generations after making suitable crosses. Before
giving some of the results obtained, it seems desirable to discuss
the principles involved.
GENETIC EXPECTATIONS FROM BACKCROSSING
As used in plant breeding, backcrossing seems to be a logical
procedure when it is advantageous to add one or two characters,
each of which is conditioned by one or two genetic factors, to an
otherwise desirable variety. The general plan of study may be
outlined as follows:
1. Selection of parents for crossing.
A variety, A, with desirable characters but lacking one or two characters
that are dependent upon only a few genetic factors.
A variety, /?, containing these one or two characters that A lacks.
2. Backcrossing of the F\ of A X B to A; selection for the one or two
desirable characters of B, if they are dominant, in each backcross gen-
eration and again backcrossing of these selected plants to A,
Repetition of the process as seems necessary. In this illustration, A
and B are called the recurrent and nonrecurrent parent, respectively.
Recessive characters of the nonrecurrent parent can be carried along
by growing sufficient plants in each backcross generation and by making
sufficient backcrosses to be sure some plants are heterozygous for the
recessive factors that it is desired to add to the recurrent parent.
3. Selection in the selfed progeny from plants carrying the factors obtained
from B until homozygosis for the characters of the B parent is reached.
101
102
METHODS OF PLANT BREEDING
In self-pollinated plants, the new lines obtained may be compared with
each other and with parent A in field trials and the strain of greatest
promise increased and distributed as an improved variety if it per-
forms satisfactorily. In cross-pollinated plants it seems necessary to
produce several desirable lines arid recombine these to produce a
synthetic variety or to use certain of these lines to produce F\ crosses
for the utilization of hybrid vigor. With a crop like com, hybrid seed
may be produced by this method. With asexually propagated crops,
the more desirable crosses may be propagated by asexual methods.
Richey (1927) has given the percentages of plants homozygous
for the n factors entering the cross only from the recurring
homozygous parent in each of r successive generations, calcu-
in Table 5.
2* \\ n
^ J
These percentages are given
TABLE 5. PERCENTAGES OF PLANTS HOMOZYGOUS FOR THE n FACTORS
ENTERING A CROSS ONLY FROM THE HOMOZYGOUS RECURRENT PARENT
TO WHICH THE ("ROSS AND THE RESULTING PROGENIES ARE
MATED IN EACH OF r SUCCESSIVE GENERATIONS*
Number of
Number of generations of back pollinating, r
lautui pa-lie ,
n
1
2
3
4
5
6
7
8
9
10
It
50
75
88
94
97
98
99
100
100-
100-
5
3
24
51
72
85
92
96
98
99
100-
10
6
26
52
73
85
92
96
98
99
15
1
13
38
62
79
89
94
97
99
20
7
28
53
73
85
92
96
98
30
2
14
39
62
79
89
94
97
40
8
28
53
73
86
92
96
50
4
20
46
68
82
91
95
75
9
31
56
75
86
93
100
4
21
46
68
82
91
* Subject to eirors incident to the use of 6-place logarithms.
t The values f or n = 1 give also the percentages of homozygous factor pairs in the entire
population, regardless of the value of n.
This formula is the same as that used for finding the percentage
of homozygous individuals in different segregating generations
after a cross. In the segregating generations after a cross, only
one-half of the homozygous individuals are of the desired geno-
type. For example, the F% generation of the cross between A A
and aa will consist of 1AA + 2Aa + 100. One-half of the
THE BACKCROSS METHOD OF PLANT BREEDING 103
progeny are homozygous, but of these only one-half are of the A A
genotype. If, instead of selfing the FI, Aa, it is backcrossed to
AA, IAA to lAa will be obtained. In this case, one-half of the
total progeny are of the desired genotype A A.
Richey gave a summary also of the number of plants required
in F and the first backcross generation to obtain a single indi-
vidual with the required genotype when one to eight factor pairs
were involved. These results, which are calculated on the basis
of independent inheritance, are given in Table 6.
TABLE 6. PROGENY REQUIRED TO HATE ONE DOMINANT HOMOZYGQUS
INDIVIDUAL
Number of factor pairs
A/T "k ,-!
1
2
3
4
5
6
7
8
FI, selfed
4
16
64
256
1024
4096
16,384
65,536
FI, backcrossed to homo-
zygous dominant
2
4
8
16
32
64
128
256
With a difference of five factor pairs in the parents, for example,
the calculated expectation in F z is only 1 individual out of every
1024, with all 5 factor pairs in a dominant homozygous condition,
whereas for the first backcross generation the theoretical expecta-
tion is 1 out of every 32 that contain all 5 factor pairs in a
dominant homozygous condition.
Linkage may be involved between a factor C for one of the
desirable characters of the recurring parent A and the recessive
condition of the dominant factor R carried in parent B that it is
hoped to add to A. Suppose, in addition, that there may be 10
other factor pairs involved in which the A parent carries the
desired genotype.
Parent A carries CC linked with rr and 10 other dominant fac-
tors, and parent B carries cc linked with RR, the latter being the
character desired to add to parent A. It may be supposed that C
and r show 10 per cent recombination. Selection for R in the
backcross generations will tend to make it difficult to obtain the
desirable factor C, but as C is brought in from the recurring parent
in each backcross the chances of a crossover and the desired
104 METHODS OF PLANT BREEDING
combination of CCRr seems better in backcrosses than by the
pedigree plan. The reason for this is that the Rr genotypes are
selected each year and the C factor brought in from the recurring
parent. After a crossover takes place, then linkage of C and R
will tend to make these combinations more frequent than under a
system of independent inheritance of these 2 factor pairs. The
10 remaining dominant factors will be recovered according to the
usual theoretical expectation.
The value of the backcross method may be better appreciated
by giving a few illustrations.
CANTALOUPES RESISTANT TO POWDERY MILDEW,
ERYSIPHE CICHORACEARUM
This disease, according to Jones (1932), cannot be controlled
satisfactorily by spraying or dusting. A mixed lot of seed from
India produced numerous plants practically free from disease,
but the melons were of very low quality. These conditions gave
an ideal setup for a backcross plan of breeding, for resistance
appeared to be a simple dominant over susceptibility, and desir-
able melons lacking resistance to powdery mildew were available.
The method used consisted of crossing the resistance Hindu
melon with commercial cantaloupes, selecting for resistance in the
backcross generations, and recrossing these selected resistant
plants to commercial cantaloupes. After cantaloupes of desired
quality were obtained, selection was followed until strains were
obtained that were homozygous for resistance. Seed from several
strains was combined to give the necessary vigor of growth in the
new variety. This material should be as uniform for other char-
acters as the original variety.
BREEDING BUNT-RESISTANT WHEATS
Briggs (1930) described a project that was started in 1922 fo:-
adding the bunt resistance of Martin to important commercial
varieties of wheat grown in California, Martin being selected for
one parent because it had proved completely resistant to bunt on
the Pacific coast and because there appeared to be only one factor
pair involved in this type of resistance. Subsequent discussions
of the backcross method by Briggs (1935, 1938) emphasize the
extensive use of the method by Briggs and coworkers.
THE BACKCROSS METHOD OF PLANT BREEDING 105
The plan outlined originally by Briggs is as follows:
1922. Martin (resistant variety) X Baart (commercial, susceptible
variety)
1923. Fi X Baart
1924. Plants segregated 1 resistant: 1 susceptible
1 925. Plants segregated 3 resistant : 1 susceptible
1926. Progeny of resistant plants segregated giving 1 homozygous:
2 heterozygous rows
Homozygous resistant plants were crossed with Baart
1927. Fi X Baait
It will be noted that the progeny of the first backcross segre-
gated in a ratio of 1 resistant: 1 susceptible, showing resistance
to be dominant. The progeny of the resistant plants of the back-
cross segregated in a ratio of 3 resistant: 1 susceptible, and the
progeny of these resistant plants gave 1 homozygous resistant
line : 2 heterozygous, on the average.
The resistant line was then used for the second backcross. The
reason for studying the progeny of selfed lines after the first
backcross, until homozygosis is again obtained, is to eliminate
the possibility of selecting a plant for backcrossing that did not
carry factors for resistance, since some plants escaped infection
even though genotypically susceptible.
The practical accomplishments of Briggs, 1 in California, demon-
strate the value of the backcross method. In 1922, a program
was started to incorporate the bunt resistance of Martin in all
commercial wheats grown in California, including the varieties
Baart and White Federation. In 1930, a program was initiated
to add the stem-rust resistance obtained from Hope to Baart and
White Federation in addition to bunt resistance. By crossing
the bunt-resistant Baart and White Federation with rust-resist-
ant strains of the two varieties, resistances to both diseases were
combined. From these studies, 11 bunt-resistant varieties, two
of these being resistant also to stem rust, have been produced.
The first group of varieties obtained from the program have been
grown extensively. The improved varieties are practically
identical with the original varieties except for the character of
bunt resistance.
A program has been adopted by Briggs and his coworkers of
compositing 70 or more backcross lines for each variety. These
1 Unpublished information kindly furnished by Fred N. Briggs.
106 METHODS OF PLANT BREEDING
lines retain the name of their original commercial type, with the
year of increase appended to designate them from their susceptible
counterpart. It has been found that the mean yield of all lines
is almost exactly the same as that of the original parent when
bunt is not a factor in yielding ability.
FIG. 13. -Top row: (A) Baart 38; (B) White Federation 38; (C) Big Club 37;
CD) Poso 41; (#) Sonora 37; (F) Pacific Bluestem 37; (G) Ramona 41; (H)
Bunyip 41; (/) Escoridido 41; (,/) Federation 41; and (K) Onas 41, all carrying
the Martin factor for resistance to bunt. Tilletia tritici and the first two, Baart 38
and White Federation 38, also have resistance to stem rust, Puccinia graminis
tritici, from Hope. The corresponding heads in the bottom row are from the
original susceptible parents.
BREEDING RUST-RESISTANT SNAPDRAGONS
Emsweller and Jones (1934) have described the development of
varieties of the cultivated snapdragon resistant to rust, Puccinia
antirrhini D. & IL, and emphasized the extent to which the
disease has become of commercial importance. From the progeny
of seed obtained originally from E. B. Mains, of Indiana, several
plants were found that under favorable conditions for infection
proved entirely free from rust. Resistance was found to be
dominant in crosses and resistance and susceptibility to be con-
ditioned by a single main factor pair, although minor modifying
factors that influenced the extent of resistance were present also.
THE BACKCROSS METHOD OF PLANT BREEDING 107
These writers describe their experiments in which they used the
backcross method to combine this resistance with the flower color
and plant habit of standard commercial varieties and state that
progress was very encouraging.
STUDIES AT THE MINNESOTA STATION
For several years, backcross studies have been carried on at
Minnesota, where the problems involved appeared to be of such a
nature that the backcross method seems the most logical plan of
breeding. Several of these problems will be outlined briefly to
give further illustrations of the principles involved.
Disease Resistance in Wheat. In spring and winter wheat,
desirable commercial varieties are available that are of good
agronomic habit and of high milling and baking quality. These
varieties lack one or two characters of outstanding importance,
notably the high resistance of Hope and H44 wheats to stem and
leaf rusts. Since resistance to both diseases is relatively simple
in inheritance, the addition of these types of resistance to desir-
able varieties available can be most logically and easily accom-
plished by the backcross method of breeding.
Its use may be illustrated by the improvement of Thatcher
through crossing with Hope, primarily to add leaf -rust resistance
from Hope and increase stem-rust resistance. Thatcher is the
standard for yielding ability, desirable agronomic characters,
and milling and baking value. It is not as stem-rust resistant as
Hope and is very susceptible to leaf rust. The following informa-
tion was furnished by E. R. Ausemus, of the Division of
Cereal Crops and Diseases, IT. S. Department of Agriculture, who
is stationed at University Farm, St. Paul, Minnesota, and who has
charge of the wheat-breeding program.
HISTOEY OF BACKCROSS (THATCHER X HOPE) X THATCHER
Year
Plan
Place
1930
Original cross
Field
1930-1931
First backcross
Greenhouse
1931
Second backcross
Field
1932-1937
Pedigree selection
Field
Leaf-rust epidemics were obtained only in 1932 and 1935.
Stem-rust epidemics were obtained each year.
108
METHODS OF PLANT BREEDING
Yield in bushels per acre, test weight, and leaf-rust reaction,
from trials made in 1938, when there was a severe rust epidemic
in the northwest spring- wheat area, are given in Table 7.
TABLE 7. YIELD IN BUSHELS PER ACRE AND TEST WEIGHT IN REPLICATED
ROD-ROW TRIALS AT UNIVERSITY FARM, 1938. LEAF-RUST REACTION
IN ROD Rows (AGRONOMY) AND IN THE RUST NURSERY (R.N.)
Rust, p
>er cent
Variety
Yield,
bu.
Test
weight
Le
af
Ste
m
Agroii.
R.N.
Agron.
R.N.
Thatcher
18
47
70
80
T
3
B.C., II-31-2
31
55
2
T
T
T
B.C., II-31-6
30
54
T
5
T
T
B.C., 11-31-14
32
54
3
T
T
T
Illustrations with Corn. The backcross method seems well
adapted to use with cross-pollinated crops that are being bred by
controlled pollination and selection. In one instance, two inbred
lines of Crosby sweet corn have been selected that combine well
together to give a vigorous F\ cross. In common with many
strains of Crosby sweet corn, these lines have the undesirable
characteristic of toughness of pericarp. Two linen of Golden
Bantam have been selected that ex( el in flavor and tenderness of
pericarp. One of these has been crossed with one of the strains
of Crosby and the other Golden Bantam line with the second
strain of Crosby.
The FI generation is intermediate in tenderness and can be
differentiated from the tough pericarp parent by puncture tests
when the canning stage is reached. This can be accomplished by
stripping back the husk at this stage and puncturing several
kernels in the middle of the ear and recording the pressure
required. The backcross method has been used in this problem.
It is not known how many genes are concerned in tenderness of
pericarp. Inbred lines are available that give considerable
ranges in mean values when measured by the puncture test. The
results in the following summary show that by selection it is
possible to differentiate heterozygous ears, by means of puncture
tests, from the homozygous tough parent. Data given in Table 8
are summarized from the studies of Johnson and Hayes (1938).
THE BACKCROSS METHOD OF PLANT BREEDING 109
The major factor or factors for tenderness was retained in the
heterozygous condition by selection in each of the succeeding
generations of backcrossing. Plants of the first backcross genera-
tion, (I X H)I, were pollinated by pollen from the I inbred
parent, tested for puncture-test values, and those that were
intermediate for tenderness were selected as parents for the next
backcross generation. After three generations of backcrossing,
selection in self-pollinated lines was used to isolate homozygous
tender pericarp inbreds that resembled the recurrent tough peri-
carp parent in most other characters,
TABLE 8. FREQUENCY DISTRIBUTION OF PUNCTURE-TEST VALUES OP EARS
FROM INDIVIDUAL PLANTS OF THE PROGENY OF BACKCROSSES TO THE
TOUGH PERICARP PARENT WITH SELECTION FOR TENDERNESS
Puncture-test values
f^nltiir^
"V
240
260
280
300
320
340
360
380
400
420
I
1935
11
40
37
4
I
1936
11
46
43
I
1937
2
16
8
H
1936
14
28
33
6
1
H
1937
1
19
16
10
2
(I X H)I
1935
4
12
23
23
15
4
1
(I X H)I 2
1936
5
14
36
40
26
13
2
(I X H)I 3
1937
2
28
43
65
40
15
6
Recent studies of inheritance of smut reaction in corn show that
the character is relatively complex from the genetic standpoint.
Attempts have been made to gain some idea of the number of
factors involved in crosses of resistant inbred lines with highly
susceptible lines that have known interchanges for certain
chromosomes. The interchange plants are semisterile, and the
point of interchange may be handled in the same manner as a
dominant factor. To determine what parts of the chromosome
map are involved in relation to factors for resistance and suscepti-
bility to smut, studies were made of smut reaction in relation to
points of interchange. Studies, at Minnesota of crosses of sus-
ceptible lines carrying interchanges with two different selfed lines
resistant to smut, one from Minn. 13, and the other from
110 METHODS OF PLANT BREEDING
Rustler, were carried on. The FI plants were crossed to the
resistant parents and in backcross and F% generations the X 2 test
for independence was used to determine associations between
smut reaction and points of interchange. In each series of
crosses, at least three different chromosomal regions carried
inherited factors for reaction to smut, and the regions from the
Rustler crosses were entirely different than those for the Minn.
13 crosses. Similar results have been reported by Burnham
and Cartledge (1939).
Various workers have found it relatively easy, however, to
select inbred lines resistai^t to smut from crosses between resistant
and susceptible inbreds and from selection in self -pollinated lines
from commercial varieties. Resistance seems to be relative and
probably functions under normal conditions against all physi-
ologic races of smut. The inbred line B164 is used as a male
parent in producing Minhybrid 301 and also in Pioneer 355, two
three-way crosses adapted to southern Minnesota. As grown in
Minnesota, B164 is highly susceptible to smut, and under normal
conditions as high as 90 per cent of the plants may be infected.
New lines, resembling B164, have been obtained from a cross
between B164 and culture 37, a resistant line of Minn. 23. Two
backcrosses to B164, followed by 3 years of self-pollination,
isolated several inbred lines that had only 10 per cent of smut
when grown adjacent to B164 that showed 85 to 90 per cent smut.
Another illustration may be given with a corn problem now
being investigated. Seed from one of the double crosses grown
commercially in Minnesota, known as Minhybrid 401, is of mixed
color, carrying both yellow and white kernels on the same ear-
It was obtained by crossing the F\ cross of two inbred lines of
yellow endosperm corn obtained from Minn. 13, lines 11 and 14,
with two white endosperm inbr : ed lines of Rustler, 15 and 19.
t It was desired to change the color of the white endosperm lines
of Rustler from white to yellow without changing their combining
ability. These Rustler lines carry the dominant whitecap factor
We, which, in the presence of yellow endosperm, causes whitecap.
The yellow lines lack this factor but contain a dominant factor
for yellow endosperm color Y. The expected results for several
backcrosses will be given on the basis of independent inheritance
of Wcwc and Yy,
Unrelated yellow endosperm lines were selected to cross with
these white endosperm lines, the problem being to obtain yellow
THE BACKCROSS METHOD OF PLANT BREEDING 111
endosperm lines that in other characters resemble the white lines
of Rustler used in the double crosses. During the backcrossing
period, selected plants in each backcross generation were crossed
with particular Rustler inbreds that were used as the recurrent
parent. In each backcross generation, the gametes of the non-
recurrent parent will be given, and the percentage of seeds
heterozygous for We and Y will be given also.
In each backcross generation, whitecap, yellow-endosperm
kernels were selected. Some of these will be homozygous for
whitecap and will be of no value. Others will be heterozygous
for both the whitecap and yellow-endosperm factors. These are
the combinations desired, and the proportions of such combina-
tions are given for each backcross generation.
1. Parent genotypes, WcWcyy and wcwcYY.
Fi genotype WcwcYy*
FI gametes WcY, wcY, Wcy, wcy.
2. First backcross genotypes and phenotypes.
a. 1 WcWcYy whitecap, yellow endosperm.
1 WcwcYy whitecap, yellow endosperm.
1 WcWcyy white endosperm.
1 Wcwcyy white endosperm.
b. Per cent heterozygous for F, 50.
Per cent heterozygous for We, 50.
Per cent heterozygous for We and F, 25.
c. Select to backcross to the Rustler inbred line.
Genotypes: WcWcYy, WcwcYy.
Gametes: 3 WcY, 1 wcY, 3 Wcy, 1 wcy.
3. Second backcross genotypes and phenotypes.
a. 3 WcWcYy whitecap, yellow endosperm.
1 WcwcYy whitecap, yellow endosperm.
3 WcWcyy white endosperm.
1 Wcwcyy white endosperm.
6. Per cent heterozygous for F, 50.
Per cent heterozygous for We, 25.
Per cent heterozygous for F and We, 12.5.
c. Select to backcross to the Rustler inbred line.
Genotypes: 3 WcWcYyj 1 WcwcYy.
Gametes: 7 WcF, 1 wcY, 7 Wcy, 1 wcy.
4. Third backcross genotypes and phenotypes.
a. 7 WcWcYy whitecap, yellow endosperm.
1 WcwcYy whitecap, yellow endosperm.
7 WcWcyy white endosperm.
1 Wcwcyy white endosperm.
6. Per cent heterozygous for F, 50.
Per cent heterozygous for TFc, 12.5.
Per cent heterozygous for F and TFc, 6.25.
112 METHODS OF PLANT BREEDING
In most problems of this kind, three backcrosses will probably
be all that are necessary. At the end of this period, in the
absence of linkage, seven-eighths of the genotype of the Rustler
inbred line will be recovered, according to theory, and if the
whitecap, yellow-endosperm seeds are selected, one out of every
eight will be heterozygous for both We and F. When planted
and self-pollinated, segregation will occur for both We and F.
From selfed ears, yellow seeds that do not carry the dominant
whitecap factor should be planted the following year, and of these
one plant out of every three, on the average, when self-pollinated,
will breed true for yellow color. In backcross studies with inbred
lines of corn, it is generally agreed that after three backcrosses the
progeny very closely resemble the recurrent parent in general
habit of growth.
From the breeding results with corn, some general conclusions
may be emphasized. When it is desired to add certain definite
characters to an otherwise desirable inbred, selection may be
practiced for these characters during the generations of backcross-
ing when the character is a dominant one. Backcrossing is
followed by selection in self-pollinated lines. If the character is
recessive, selection may be made after each backcross by self-
pollination and selection for the homozygous recessive before
making further backcrosses, or selection may be made for the
recessive character obtained from the nonrecurrent parent during
the segregating generations when selfing is practiced.
In certain corn crosses, one of the selfed parents excelled in
most easily observed characters, such as resistance to smut, good
root system, and ability to withstand lodging and in general
vigor. The other parent was less desirable in most easily
observed characters. The FI crosses were vigorous and far
superior to either of their inbred parents. It seemed relatively
easy by backcrossing and selection to obtain inbred lines with
marked improvement over the more undesirable parent but
rather difficult to obtain lines better than the more desirable
inbred from backcrosses to this inbred as the recurrent parent.
A backcross program may be most advantageously pursued when
definite characters may be selected for. The problem seems
more difficult in selecting for such characters as vigor of growth
that a^e dependent upon the interaction of many factors.
CHAPTER VIII
BREEDING FOR DISEASE AND INSECT RESISTANCE
The principles underlying the breeding for resistance to disease
or insect attacks are much the same as for other characters.
There is, however, one important difference. In breeding for
disease and insect resistance, one is dealing with two series of
heritable factors: (1) the heritable characters of the host plant
and (2) the heritable differences in the organism.
In most cases, the plant breeder is interested primarily in the
reaction of selected strains, varieties, and hybrids under condi-
tions to which they will be exposed normally when grown for
economic purposes. The breeding program should be carried on
under conditions as similar to those to be encountered by the
commercial varieties as is feasible.
It is in breeding for resistance to diseases that the modern
plant breeder has made some of his greatest contributions. The
development of special methods for producing artificially induced
epiphytotics and fundamental studies of the genetics of the host
and parasite have laid the foundations for the planned breed-
ing programs of the present. This fundamental information,
although coming slowly in the beginning, has increased rapidly in
recent years, leading to a scientific appreciation of the problems
and, in many instances, to a sound basis for solution.
THE IMPORTANCE OF DISEASE RESISTANCE
The development and utilization of disease-resistant varieties
is one of the important methods of disease control. When the
required resistance can be obtained in combination with other
necessary qualities, it seems fair to conclude that the growing of
disease-resistant varieties is the most desirable method of con-
trolling diseases. Much has been accomplished already, but
there is an almost unlimited opportunity for further work.
Seasonal variations in climatic factors are the major causes of
seasonal variations in the yielding ability of crops, and to a
considerable extent these variations cannot be controlled. The
113
114 METHODS OF PLANT BREEDING
development of disease-resistant varieties will help to stabilize
yields, since disease epidemics are among the more important
causes of wide fluctuations in the yield of crops from season to
season.
With disease resistance, as with other characters, it is important
to use as large numbers as possible in the breeding program,
although the source of material is of greater importance. A good
illustration of these two points is the present plan for breeding
high-quality European grapes that are resistant to Phylloxera,
the vine louse, and Peronospora, the vine mildew. According to
Baur (1931) the cost of attempting to control these pests amounts
to between 30 and 50 million marks annually. American varie-
ties of grapes, Vitis rupestris, are resistant to both Phylloxera and
Peronospora. The American grapes are of low quality, whereas
the European varieties, V. vitifera, are of high quality but
susceptible. Baur stated that crosses between these two species
are fertile, and he had under way large-scale experiments to
select, from the segregates of hybrids between the two species,
varieties that excel in both quality and resistance. The tests for
mildew resistance are made at the plant-breeding station at
Miincheberg, and those for resistance to vine louse at the
Institute for Phylloxera Research at Naumberg. In another
publication (Baur 1933), the statement has been made that from
5 to 7 million F% seedlings are grown yearly. They are tested for
reaction to mildew, and only the seedlings resistant to mildew are
saved and tested for yield and quality. Those that survive are
then tested for resistance to the vine louse.
Flax is grown as a cash crop in the United States in the spring-
wheat region of the central northwest, largely for the oil content
in the seed. "In earlier years, it was grown chiefly on new break-
ing, i.e., prairie soil not previously under cultivation. Eventu-
ally no new soil was available, and it became necessary to grow
the crop on old land. Serious losses from wilt occurred, and it
seems very probable that flax could not have continued to be a
successful crop without the development of resistant varieties,
In 1901, Bolley grew normal varieties on " wilt-sick soil" and
isolated the organism Fusarium lini responsible for the disease.
He observed (1901) the fact that some plants were not seriously
injured even under epidemic conditions. These and other studies
continued until the present time have emphasized the value of
BREEDING FOR DISEASE AND INSECT RESISTANCE 115
wilt resistance. Today no improved varieties of flax are intro-
duced unless they have the necessary resistance to wilt.
In more recent times, wilt-resistant varieties have been devel-
oped for many plants. In 1915, cabbage growers in Wisconsin
were 'so discouraged because of the ravages of cabbage yellows
(F. conglutinans) that they were about to abandon cabbage grow-
ing. Jones and Gilman (1915) observed a few plants in fields
infected with the disease that apparently escaped the disease.
These were selected and proved to be resistant by progeny trial.
Today cabbage yellows is no longer serious. Wilt-resistant
strains of all important varieties grown in Wisconsin have been
developed.
By similar methods, wilt-resistant varieties of tomatoes have
been obtained. Edgerton (1918) and others, in Louisiana, have
developed an improved technic for selection. Seed was planted
in sterilized soil and then inoculated with a pure culture of the
wilt-producing organism F. lycopersici. Those seedlings that
were injured by wilt were pulled up and discarded and the resist-
ant seedlings transplanted to a field that was known to contain
infected soil. Lines breeding true for resistance were tested for
other characters and distributed when satisfactory.
Orton (1913) developed watermelons resistant to wilt, caused
by F. niveum, by crossing an inedible, resistant citron with a good-
quality, susceptible watermelon and selecting for resistance and
quality.
About 1920, farmers in western New York were greatly con-
cerned over the bean-mosaic disease. Emerson and coworkers
had learned that a variety known as M. A. C. Robust, devel-
oped by selection in Michigan, was resistant to mosaic. They
introduced the new variety, planting single rows in the center of
the farmer's fields, and were greatly pleased with the great
resistance that this variety exhibited. To quote from Emerson,
"I have no hesitation in saying that resistance to this disease,
mosaic, saved the pea-bean industry in Western New York."
Some of the accomplishments during the last 25 years in the
production of disease-resistant varieties of vegetables have been
summarized by Rieman (1939). A study of lists of varieties of
vegetables offered for sale by the seed trade was made in 1914.
At that time, less than a dozen resistant varieties were listed by
two of the leading American vegetable seed houses, and most
116 METHODS OF PLANT BREEDING
of these proved of doubtful value. In 1939, over 80 resistant
varieties were listed, and 20 or more were recognized by the seed
trade as leading varieties. These 80 disease-resistant varieties
included 2 varieties of asparagus resistant to rust; 3 varieties of
snap-bean resistant to mosaic and 2 resistant to bean rust; 10
varieties of cabbage resistant to cabbage yellows; 1 variety of
celery resistant to Fusarium wilt; 6 varieties of sweet corn resist-
ant to Stewart's bacterial wilt and 2 resistant to the corn-ear
worm; 9 varieties of lettuce resistant to brown blight, 3 resistant
to downy mildew, and 2 resistant to tip burn; 2 varieties of
cantaloupe resistant to powdery mildew; 29 varieties of peas
resistant to Fusarium wilt; 2 varieties of spinach resistant to
mosaic; and 7 varieties of tomatoes resistant to Fusarium wilt.
Rieman stated that the method of control through breeding
disease-resistant varieties involved four important steps: (1)
the recognition of disease symptoms and the identification of the
causal organism, (2) the isolation of fertile, resistant breeding
stocks, (3) the development of true-breeding disease-resistant
varieties through crossing and selection, and (4) the production
and distribution of pure high-quality seeds of the resistant varie-
ties in commercial quantities.
METHODS OF BREEDING FOR DISEASE AND INSECT RESISTANCE
The methods of breeding will be discussed under several head-
ings, namely: (1) the search for resistant materials, (2) the artifi-
cial production of the disease epidemic, (3) the plan of breeding,
and (4) a study of fundamental problems that aid in a logical
attack on the breeding problem.
THE SEAECH FOR RESISTANT MATERIAL
The search for resistant varieties or strains is a logical first step.
It is highly desirable, in beginning an attack on a disease problem,
to make a collection of local and foreign varieties and to test their
reaction under epidemic conditions. If reaction to the disease
has been studied by investigators in other states or other coun-
tries, varieties found by them to be resistant would, of course,
be tested first to determine their reaction under the environ-
mental conditions of the new investigation.
The methods of breeding varieties with the necessary resistance
do not differ from those used for other characters. In some
BREEDING FOR DISEASE AND INSECT RESISTANCE 117
cases, hybrids between susceptible varieties may give resistant
plants in the segregating generations. If disease resistance is an
important problem, at least one of the parents of a hybrid should
have the desired resistance whenever such resistant varieties
are available.
With disease resistance, as with other characters, it is extremely
important to learn as much as possible regarding the genetic
factors responsible for resistance. It must be appreciated that
the organism causing the disease frequently comprises several (or
many) strains or physiologic races, which in some cases can be
differentiated only by their manner of reaction to a series of
varieties known as differential hosts. Almost 200 such phys-
iologic races have been found in Puccinia graminis tritici.
For stem rust of oats (P. graminis avenae), there is a complete
correlation between the manner of reaction of seedlings and the
reaction of the mature plants to the same physiologic races (Smith
1934). In barley, Brookins (1940) found the reaction of mature
plants to a large collection of physiologic races in the field to be
controlled by the same genetic factor pair controlling reaction to
races 19, 36, and 56 in the seedling stage. Resistance to stem
rust in wheat may be placed in two main classes: (1) physiological,
where the reaction to a specific race of the pathogen is relatively
the same throughout the life of the plant, and (2) mature-plant
resistance, in which case the genes concerned produce resistant
mature plants independently of their reaction in the seedling
stage.
It is necessary in most cases to make a survey of the physio-
logic races normally prevalent in the locality and to breed for
resistance to all prevalent forms. This is essential in dealing
with physiological resistance, since a variety may be very resist-
ant to one physiologic race and highly susceptible to another.
With mature-plant resistance to stem rust of wheat and smut in
corn, varieties tend to react in the same manner to all physio-
logic races of the pathogen. In studies of disease resistance, it is
very helpful to have the definite cooperation of the plant pathol-
ogist, who will lead in the studies of the organism and in
methods of creating artificial epidemics. After -these problems
have been solved, the breeding of disease-resistant varieties
is not greatly different from that of breeding for any other
character.
118 METHODS OF PLANT BREEDING
ARTIFICIAL PRODUCTION OF EPIPHYTOTICS
It is not within the scope of this book to describe methods of
inducing disease or insect epidemics for all crops in which breed-
ing for resistance is being undertaken. A brief description will
be given for some of the major diseases of the common field crops
grown in the northern United States and Canada. These will be
described for the crop and disease, or insect pest, involved.
Black-stem rust of wheat, oats, and barley (Puccinia graminis).
1. Field.
a. Plant susceptible varieties as borders around the outside of the plot
and through the alleys.
b. Obtain as many physiologic races as possible that have been found
in the region . Increase these races on seedlings of susceptible varieties
in the greenhouse, arid use a mixture of races for inoculations in the
field.
c. Transfer rusted plants from the greenhouse to the border rows in the
field.
d. Hypodermically inoculate the border rows with a mixture of uredo-
spores of all the races increased in the greenhouse. A mixture of
about 30 races has been used at Minnesota in recent years.
e. Spray the plants with an aqueous spore suspension late in the evening,
when there is probability of dew, or just before or after a rain.
/. Keep the soil moist in a dry season to delay premature ripening and
prolong the length of the susceptible period.
2. Greenhouse.
When there is correlation between seedling reaction in the greenhouse and
reaction in the field from heading to maturity, it is of advantage to study
the progeny of selected plants in the greenhouse as an aid in discarding
susceptible material,
a. Grow from 15 to 20 seedlings from selected plants in small pots until
first leaves are well developed.
6. Spray seedlings with water, and inoculate by brushing with leaves of
infected seedlings, or apply the spores with a scalpel.
c. Place the pots in an incubation chamber under high humidity for
48 hr. The chamber may have a glass top to admit light.
d. Transfer pots to greenhouse bench, and observe the reaction when
rust has developed to the point where maximum differentiation is
obtained.
Leaf rust of wheat (Puccinia triticina] and crown rust of oats (P. coronata).
1. Increase rust of the races to be used on seedlings in the greenhouse.
2. Plant susceptible varieties as border rows around and through the field-
rust nursery.
BREEDING FOR DISEASE AND INSECT RESISTANCE 119
3. Spray the plots with an aqueous suspension of all races usually prevalent
' in the locality. The plants should be inoculated on a still night, when
the humidity is high. Seedlings may be inoculated in the field when
about 8 in. high.
4. Irrigate, if necessary, to maintain susceptibility over a period of time.
5. Tag resistant plants, if the lines are segregating. Final selections are
made at harvest time from these resistant plants.
__
FIG. 14. Hypodermic inoculation of border rows of susceptible plants with
a spore suspension of a collection of races of stem rust.
Bunt of wheat (Tilletia tritici).
1. Obtain as many collections of smut as possible from a wide area in
order to obtain a large number of races. Do not use collections of smut
from too wide an area or from foreign countries, for there is danger in
introducing more virulent forms of the pathogen.
2. Dust the seed of varieties or hybrids to be tested with a mixture of spores
from all the collections, using about 1 g. of smut per 100 cc. of grain.
3. Plant the seed as early as possible, if spring wheat, or when the soil is
sufficiently cool and relatively dry. The optimum temperature for
infection is approximately 12G,
120 METHODS OF PLANT BREEDING
Loose smut of oats (Ustilago avenae), covered smut of oats (U. levis), covered
smut of barley (U. hordei), and intermediate smut of barley (U. medians).
1. Obtain as many collections as possible in the area in which the variety
may be grown ultimately.
2. Make an aqueous suspension of the spores at the rate of J/ g. of spores
to 100 cc. of water.
3. Submerge seed in about one and one-half to two times its volume of the
spore suspension.
4. Subject to sufficiently high vacuum to withdraw air from under the
hulls. Two evacuations in succession are preferable.
5. Pour off the suspension., and dry the seed.
6. The seed may be stored for several weeks before planting without greatly
affecting the efficiency of inoculation.
7. Plant when temperature is moderately high and the soil relatively dry.
Flax rust (Melampsora lini).
1. Increase the rust in the greenhouse.
2. Plant border rows of susceptible varieties around and through the field
nursery.
3. Spray a water suspension of spores on border rows of susceptible varieties.
When rust appears, make spore suspensions, arid spray the rust over the
entire nursery, or brush the plots with infected plants from the border
rows.
4. Save bundles of rusted flax plants in the fall, and scatter the straw on the
plots in the spring, when the plants are coming up.
5. Grow the flax in a place where the temperature is relatively low and
humidity is high, since these conditions are conducive to the develop-
ment of an epidemic.
Flax wilt (Fusarium lini).
1 . In the greenhouse.
a. Plant susceptible varieties to be tested in sterile soil inoculated
with the cultures of the causal organism.
2. In the field.
a. Collect soil from fields where flax wilt has been prevalent, and mix
this with the soil in the test plot.
b. Grow a mixture of races of the causal organism on sterile grain,
nutrient agar, or in liquid media, and inoculate the soil.
c. Plant the varieties and strains of flax to be tested in this "wilt
nursery."
d. Use the same plot every yealr.
Fusarial head blight (scab) of wheat and barley.
1. Increase the different organisms on sterile oats or wheat in jars or
flasks in the laboratory.
2. At time of heading, cover the field-test plot with a cloth tent.
BREEDING FOR DISEASE AND INSECT RESISTANCE 121
3. After heading, spray the plants every day or two with a spore suspension
made from the different organisms. Continue the spraying until the
grain is in the soft-dough stage, or until a satisfactory epidemic has
been produced.
4. Spray the plants, soil, and tent with water to maintain high humidity in
the tent.
FIG. 15. Flax wilt nursery showing resistant and susceptible strains, the latter
entirely killed by wilt.
Corn smut (Ustilago zeae).
1. Make as many collections of smut as possible in the region in which the
corn may be grown,
2. Mix the collections of chlamydospores with manure, and spread between
the rows when the corn is 6 to 12 in. high.
3. Dust chlamydospores on the plants, or spray with a spore suspension,
three or four times during the growing period of the com.
4. Use the same field as a "smut nursery' 7 every year.
Loose smut of sorghum (Sphacelotheca cruenta) and covered smut of sorghum
(S. sorghi).
1. Make collections the previous year from as many sources as possible,
2. Dust the spores on the seed before planting.
122 METHODS OF PLANT BREEDING
3. Plant the seed when fairly high temperatures are assured. The optimum
temperature for infection is about 27C., and very little infection will
result below 15C.
Head smut of sorghum and corn (Sorosporium reilianum).
1. Make as many collections of the organism as possible.
2. Mix chlamydospores with soil, and spread this over the field-test plot, or
apply with seed at planting time.
3. Plant in relatively dry soil.
4. Use the same plot every year.
Hessian- fly injury of wheat.
1. Plant replicated Hessian-fly nurseries in short rows, with check rows of
varieties of known reaction at regular intervals.
2. Import infested stubble from the localities in which the wheat is to be
grown ultimately. "Flaxsecd," secured by dissecting plants from pre-
ceding tests, may be used also. Place these between the rows, and
sprinkle with water.
3. Several biologic or physiologic races of the fly are known. The races
produce a different reaction on different varieties of wheat.
4. In the greenhouse, grow the plants in pots, and transfer to an enclosure
in the insectary after the plants are well tillered. Allow adult flies to
lay eggs, and after emergence of larvae, return the pots to the regular
greenhouse. After the fly has reached the "flaxseed" stage, the plants
are dissected and the infestation recorded.
METHODS OF BREEDING
It is desirable that the early generations of selections and the
segregating generations from crosses be grown under controlled
epidemic conditions, so that the resistant strains can be selected
and the others eliminated. It is generally considered that a
mixture of races, or collections, of the pathogen, from the general
region in which the improved crop will be grown, should be used
in producing the epidemic. There is, however, some danger in
carrying inoculum from one locality to another, since a virulent
form of the disease may be introduced. For this reason, the
more promising of the selections may be grown in several locali-
ties and exposed to an epidemic in each locality, created by using
disease organisms collected in that locality.
The fact that plant pathogens are composed of physiologic
races that can be differentiated only by their manner of reaction
on a series of varieties known as differential hosts is appreciated
today by students of plant breeding. A knowledge of the number
and nature of these physiologic races is important, Two
BREEDING FOR DISEASE AND INSECT RESISTANCE 123
general methods may be followed. These are not necessarily in
opposition to each other but may go hand in hand. One consists
of a study and isolation of as many physiologic races of the
organism as are available and a determination of the reaction of
parents and hybrids to individual races. The other method
consists of using a collection of races in producing the epidemic
and selecting only those strains that are resistant to all races.
Knowing the reaction of parental varieties to special races
makes possible a definite breeding program ir_ relation to these
races. Thus, with the physiologic races of stem rust of wheat,
Puccinia graminis tritici, to which Kanred is immune both in the
seedling and in the mature-plant stage, immunity from a rather
large group of races is dependent upon a single genetic factor.
Breeding for resistance to a single one of these races gives an
accurate idea of the reaction to all races from which Kanred is
immune.
An illustration may be given from the studies of Smith (1934)
with respect to stem rust of oats.
TABLE 9. REACTION OF VARIETIES OF OATS TO NINE RACES OF STEM RUST
Variety
Reaction to races
1
2
3
4
5
6
7
V 8
White Russian
R*
R
R
S
R
R
S
S
st
R
R
S
S
S
R
S
72
R
S
S
S
S
S
S
S
R
S
S
R
S
S
S
Rainbow
Joanette ...
Victory ...
* R resistant,
t S susceptible.
Races 1, 2, and 5 are the most common in the central northwest
section of the United States. It seems probable that reaction to
races 1, 2, 3, 5, and 7 may be dependent upon an allelic series of
three factors, one governing resistance to all five races, another
resistance to races 1, 2, and 5 and susceptibility to 3 and 7, and
still another susceptibility to all five races. Rainbow, a selection
from dreen Russian, made at the North Dakota Agricultural
Experiment Station, is resistant to all five races; White Russian is
resistant to 1, 2, and 5 but susceptible to 3 and 7. The resistance
of Rainbow is somewhat greater to races 1, 2, and 5 than that of
124 METHODS OF PLANT BREEDING
White Russian, and as it is resistant also to races 3 and 7, its type
of resistance is more desirable than that of White Russian.
It is advantageous in some cases to duplicate the nursery,
making two planting dates. The material that gives the most
satisfactory disease epidemic may be used for selection and the
material from the other planting date discarded.
Breeding for disease resistance can be carried on most advan-
tageously by carrying on the studies of disease reaction as a part
of the main breeding project, selecting for disease reaction, for
quality, and for agronomic characters at the same time, although
in some cases in special nurseries. In this way, if selection must
be made for several characters, progenies that excel in all these
characters may be used as a basis of selection.
STUDY OF FUNDAMENTAL PROBLEMS
The importance of a knowledge of the pathogen and the
environmental conditions favorable for the development of the
disease will be appreciated by students with training in plant
pathology. From extensive studies with many organisms, more
especially species of rusts, smuts, powdery mildews, pasmo,
Fusarium root rots, Helminthosporiurrij Colletotrichurrij and others,
it has been learned that many and perhaps most of the organisms
causing diseases are composed of numerous physiologic races
that can be differentiated only by the manner of their reaction on
a series of varieties known as differential hosts.
The identification of physiologic races of pathogens responsi-
ble for particular diseases has made possible a more definite
attack on fundamental problems, such as the effect of environ-
mental conditions in causing marked changes in disease reaction,
the ability of different races to develop to epidemic proportions
as affected by environment, the screening effect of varieties as a
means of modifying the proportion of particular races present
during any particular season, and other similar problems. These
questions belong logically in the subject-matter field of plant
pathology and cannot be handled adequately here. An illustra-
tion of the manner of identification of physiologic races based on
host reaction will be given for loose smut of oats, Ustilago levis,
after Reed (1940). Differential species and varieties used
include A vena brevis; A. strigosa; A. sativa, varieties Black Dia-
mond, Black Mesdag, Black Norway, Danish Island, and
BREEDING FOR DISEASE AND INSECT RESISTANCE 125
Monarch; ^4. saliva orientales, variety Green Mountain; A. nuda,
variety Hull-less; A. byzantina, variety Fulghum. The manner
of differentiation is given in Table 10, taken from Reed.
TABLE 10. DIFFERENTIATION OF SPECIALIZED RACES OF Ustilago levis
(K. & S.) Magn.
a. Monarch susceptible Race
b. Black Mesdag susceptible
c. Fulghum susceptible
d. Black Norway susceptible 6
d. Black Norway resistant
e. Black Diamond susceptible 7
e. Black Diamond resistant 8
c. Fulghum resistant 9
b. Black Mesdag resistant
c. Black Norway susceptible
d. Green Mountain susceptible 13
d. Green Mountain resistant 14
c. Black Norway resistant
d. Green Mountain susceptible
e. Danish Island susceptible
/. Hull-less suscept^ble 11
/. Hull-less resistant 12
e. Danish Island resistant
/. Avena strigosa susceptible 1
/. Avena strigosa resistant. . . . , 10
d. Green Mountain resistant
e. Black Diamond susceptible 4
e. Black Diamond resistant 3
a. Monarch resistant
b. Avena brevis susceptible 2
b. Avena brevis resistant
c. Hull-less susceptible 5
A summary of the status of physiologic races has been made
by Reed (1935) and Stakman et al (1935). Studies of physi-
ologic races and the use of races common to the region where the
improved resistant varieties, when obtained, will be grown, will
aid the breeder in producing varieties with the necessary resist-
ance to races common in the locality. A knowledge of stability
of pathogenicity and the basis of new races in terms of mutation
or hybridization is essential.
The genetics of plant pathogens has been reviewed recently by
Stakman et al. (1940). A few examples will be cited to illustrate
principles involved.
126 METHODS OF PLANT BREEDING
Selfing of individual physiologic races of black-stem rust on
the barberry often leads to the isolation of several to many differ-
ent races. Hybridization between races within the same species
or between different species of rusts may lead to the production
of new races. When two races of stem rust, homozygous for
pathogenicity, are crossed, the Fi resembles one or the other
parent, and there appears to be Mendelian dominance. If
heterozygous races are crossed, new races with greater virulence
than either parent may be obtained.
A knowledge of the nature and frequency of mutation for
pathogenicity is fundamental, especially the probability of a
change from a mild form of the disease to a more virulent form.
From studies made so far, it would appear that mutations that
lead to an increase in virulence are relatively infrequent.
A knowledge of the genetics of the pathogen is very important
in planning breeding studies in which the breeding for disease
resistance is one of the major objectives. Close cooperation
between plant breeders and plant pathologists would appear to be
essential if maximum progress were to be made.
A knowledge of the nature and causes of disease resistance in
terms of physiology, morphology, or functional behavior are basic
to a real understanding of the problem.
Walker (1935) discusses the nature of resistance to cabbage
yellows and describes two categories of inherited resistance that
have been obtained. In Type A, resistance is dominant to
susceptibility and controlled by a single dominant gene. All
collections of the parasite react in the same manner to plants of
Type A, and the behavior is constant over a wide range of
temperatures. There is a second type of resistance, known as
Type B, which is complex in inheritance, and the reaction varies
with the temperature, all plants being susceptible at a soil
temperature of from 22 to 24C. Thus, plants of the Type A
resistant group may be selected by raising the soil temperatures
to 24C. While the physiological or morphological differences
that differentiate the two types of resistance are unknown, the
information regarding genetic differences makes it possible to
breed for the required resistance.
From the plant-breeding standpoint, there are two major types
of resistance to stem rust. Stakman (1914) showed that resist-
ance to certain physiologic races of Puccinia graminis tritici is
BREEDING FOR DISEASE AND INSECT RESISTANCE 127
due to physiological incompatability between the host plant and
the fungus. The germ tube of the fungus may enter resistant
varieties, but the fungus is unable to establish itself to the extent
that it can cause severe injury. Physiological or protoplasmic
resistance functions throughout the life of the plant. This
appears to be the type of resistance to stem rust found in oats
(Smith 1934) and barley (Brookins 1940).
The second general type of resistance to stem rust in wheat is
called mature-plant resistance. Some varieties are susceptible to
one or more races in the seedling stage yet resistant to these and
other races in the stage from heading to maturity. The exact
nature of mature-plant resistance is not known. Whatever its
nature, the mode of inheritance of this type of resistance appears
to be relatively simple in crosses differentiated by this type of
resistance and involving crosses with Hope or H44. It appears
to function against all physiologic races found in the spring-
wheat region of the United States and Canada.
A knowledge of the mode of inheritance of reaction to disease
helps in planning the breeding prograwn. The genetics of reaction
to disease frequently can be studied during the segregating gen-
erations of hybrids made during the course of the regular breeding
program at little additional cost. Such information, analyzed
and reported, will leave an ever-increasing store of information
that will be available as a guide to future investigations.
The extent of correlation between reaction to disease and other
important character differences should be determined. A
knowledge of genetic linkage may serve as a guide in planning
breeding programs or in fundamental inheritance studies.
Cooperation between breeders in different states and countries
will be of benefit to all. New facts will become known earlier,
and free exchange of material and ideas will speed up the solution
of all breeding problems. Jealousies leading to the withholding
of information from others working on similar problems will
impede such progress. The testing of material in many places
leads to a more rapid determination of its real value.
Painter et aL (1940) described extensive experiments in Kansas
in breeding wheat resistant to Hessian fly [Phytophaga destructor
(Say)]. From crosses of Marquillo, a Hessian-fly resistant spring
wheat, with desirable varieties of winter wheat susceptible to
Hessian-fly attacks, these workers have nroduced strains of winter
128 METHODS OF PLANT BREEDING
wheat that combined resistance to Hessian fly and tolerance to
wheat joint worm, with resistance to leaf rust, stem rust, bunt,
and mildew. The fly resistance of Marquillo is probably derived
from its lumillo durum parent. In crosses, the resistance of
Marquillo tends to be recessive and due to more than a single
genetic lac tor. Resistance appears to be due to the interaction
of three separate heritable mechanisms: low larval survival,
ability to withstand infestation, and, under some conditions, low
oviposition. The best explanation of the differences in varietal
behavior in different regions lies in the presence of biological
strains of the fly, which differ in their ability to infest different
varieties of wheat.
CHAPTER IX
INHERITANCE IN WHEAT
The relationship of Triticum and related genera was given in
Chap. II by means of genom analysis. In vulgare wheats, for
example, there are three genoms, called A, B, and C, each con-
taining a set of seven chromosomes. Thus, for many characters,
there may be three pairs of factors for the dominant and recessive
condition. If, as some believe, the basic chromosome number
of the Gramineae is five instead of seven, a greater number of
duplicate factors than three could be accounted for, depending
upon the means by which the basic number five was changed to
seven. Instances of three duplicate factors are common in
members of the Spelta group, including Triticum vulgare, T.
compactum, and T. spelta, and two duplicate factors have been
found in members of the Eminer group, including T. dicoccunij T.
durum, T. turgidum, and T. polonicum. It is well to recall the
fact that T. monococcum carries genom A; T. dicoccum, T.
durum, T. turgidum, and T. polonicum, genoms A and 5; and
T. vulgare, T. spelta, and jP. compactum, genoms A, B, and C.
Studies of glume shape help to illustrate types of inheritance
that may be expected in amphidiploids.
Glume Shape. Extensive studies of glume shape and keel
development have been made by Watkins (1940), who has sum-
marized the present status of the problem. Wheat species may
be described as follows:
Hexaploids.
vulgare round, loose glumes and tough rachis.
speltoid keeled, thick glumes and tough rachis.
speUa keeled, very thick glumes and brittle rachis.
Tetraploids.
urum
looge gimneg and.tough rachis.
*
dicoccum keeled, thick glumes and brittle rachis.
129
130 METHODS OF PLANT BREEDING
Watkins concludes that the tetraploids contain two sets of
factors, perhaps representing completely linked groups of genes
with the genetic formulas
dicoccum K d K d K d K d
turgidum KK KK
He also suggests the formulas for the hexaploids to be
vulgare kk KK K d K d
speltoid KK KK K d K d
spelta KK* KK K d K d
In crosses between turgidum and dicoccum, the FI has the
formula K d K K d K and is somewhat intermediate, resembling
dicoccum more closely than turgidum, with rather thick glumes
and intermediate brittle rachis. In the FI meiosis, autosyndesis
probably occurs as first suggested by Darlington (1927) to explain
the reason for a lack of complete recovery in the segregating gen-
erations of parental glume lengths in crosses between poloni-
cum X durum f called shift by Engledow (1920). In the cross
of turgidum X dicoccum, the FI gametes presumably are all
K d K because of pairing in FI in the form of K d K d KK, leading to a
true breeding form in F% that resembles the FI.
In crosses of vulgare with dicoccum and turgidum f respec-
tively, it was concluded that K d remained unpaired when crossed
to turgidum and K, when crossed to dicoccum^ which is in agree-
ment with the lack of pairing of K and K d in the turgidum X
dicoccum cross.
With this hypothesis, /b, K, K d , and K 8 are allelic groups of
completely linked genes, and K d and K are similar or identical
in effect. The glume, keel, and rachis characters that differenti-
ate the five wheat species are caused by variations in a single
chromosome complement, present four times in tetraploids and
six in hexaploids.
Watkins presents evidence also of a linkage between the factor
pair for bearded vs. tip awns,/ called Bibi, and for glume condi-
tion, K d k or. Kk f with a recombination value of approximately
41 per cent.
Awnedness. There are three major groups of wheats: awnless,
awnleted, and bearded. The awnleted groups produce short
awns, these being longer and more numerous, usually near the
INHERITANCE IN WHEAT 131
tip of the spike in one group of wheats, and more evenly dis-
tributed in another group. Probably as a result of minor modi-
fying factors or allelic series, there are intermediate classes also
that in some cases breed true. Homozygous awnleted varieties
may differ in the extent of development of awns, and definite
classification of the genotype of homozygous material is difficult
without a breeding test.
Watkins and Ellerton (1940) have postulated the following
factors for different types of awns in hexaploid wheats:
BI, the gene for tipped 1 belonging to the allelic series BI, 61, and bi a .
A few awn tips up to 1 to 2 cm. in length are produced, the longest
tips being found near the top of the spike.
61, recessive gene for bearded.
bi a , belonging to a series of allels including BI and 61 ; producing half-
awned types with short awns,
J5 2 , the gene for tipped 2 belonging to the allelic series B 2 , bz and perhaps
containing A. Awns reduced to a few short tips occurring from top to
bottom of the spike.
&2j recessive gene for bearded in the presence of the homozygous condition
for bi.
A, another gene for the half-awned condition that may be an allele of
the 62 series and that gives half-awned types in the presence of the
recessive condition for 61 and 62-
Hd, a gene that reduces the length of the awns, making them curved and
twisted near the base.
Possible combinations of factors include 6161 i> 2 >2 hdhd, bearded;
BiBi bj)2 hdhd, tipped 1 ; 6161 J? 2 5 2 hdhd, tipped 2; 6161 6 2 &2 HdHd,
hooded; and B\B\ B^B?, hdhd, beardless; and bibi B^B^ HdHd,
hooded beardless.
Bi is linked with genes for pubescent node, square-headedness,
and keeled glumes.
The Howards (1915), in India, explained the results of a cross
between awnless and bearded varieties on the basis of two pairs
of factors, the homozygous dominant condition of both leading
to the production of the fully bearded condition. Quisenberry
and Clark (1933) crossed two awnleted wheats, Quality and
Sonora, and obtained true breeding awned, awnleted, and awnless
wheats, respectively, in F 9 in addition to a wide range of segregat-
ing groups. They used the same hypothesis as given originally
by the Howards except that they considered awnless to be the
dominant group rather than bearded.
METHODS OP PLANT BREEDING
The student should not be confused by the question of domi-
nance. It is relatively easy to differentiate the Fi from either the
awnleted or fully bearded parents when representative varieties
are crossed that differ in awnedness. Percival (1921) reported
F% segregations approximating a 1:2:1 ratio, with the intermedi-
ate or heterozygous condition producing longer tipped awns that
often extended down the head to a greater extent than in the
awnleted parent.
' -' L V!4
- .,<"$" L.M
^,i*SP
FIG. 16. Quality and Sonora, two types of tipped awn wheats, and five
different types of progeny obtained in Ft. Two new homozygous types were
selected, awnless and bearded. (After Quisenberry and Clark, 1933.)
Chaff Characters. Color of glumes is a varietal character
ranging from deep brownish red color to colorless. Segregation
in crosses between colored vs. colorless with 3:1 and 15:1 ratios
has been reported (Biffin 1905, Kezer and Boyack 1918). There
are inherited differences in color of awns that have been reported
to segregate in simple ratios (Howards, 1915).
Hairy chaff is a varietal character used in classification. In
crosses between members of the Emmer group with varieties of
Triticum vulgare, some cases of complete association have been
reported between chaff color and the hairy chaff character
INHERITANCE IN WHEAT 133
(Biffin 1905, Engledow 1914, Henkemeyer 1915, Kezer and
Boyack 1918). The Howards have reported two kinds of hairs
on the glumes of Rivet wheat. In a cross between two Indian
varieties that differed in the sorts of hairs produced on the chaff,
a ratio of 15 pubescent: 1 smooth was obtained in F^ It seems
probable that the two pairs of factors for pubescent vs. smooth
chaff are carried in separate genoms and therefore are independ-
ently inherited. Since there are at least two pairs of factors for
pubescence, this would explain variations in linkage relations
between chaff colors and pubescence.
Seed Characters. Color of seed, resulting from a brownish
red pigment in the testa, has been commonly used in varietal
classification and in determining market grades. It is a plant
character and is not immediately affected by cross-pollination.
Red is dominant over white, and from one to three pairs of dupli-
cate factors are involved, as first shown by Nilsson Ehle (191 la)
and later found by many other workers. Segregating ratios
3:1, 15:1, and 63:1 have been observed, and crosses between two
varieties, both breeding true for red seed color, may give plants
lacking red seed color in F%, provided that the parents differ in the
genetic factors involved. One parent, for example, may be
R iRi rtfrz, whereas the other may have the genotype r\r\ RzRz,
leading to the production of some white-seeded plants of the
genetic constitution r\r\ ry^.
Texture of seed is used also in varietal classification and in
market grades. Biffin (1916) observed the immediate effect of
cross-pollination in a cross of Rivet, a corneous seeded variety
belonging to Triticum turgidunij with a soft Polish variety of
T. polonicum. In crosses of corneous-seeded duruins with the
soft-seeded variety Sonora, belonging to T. vulgar e, Freeman
(1918) observed variation in texture of seed in FI with hard,
intermediate, and soft-seeded kernels on the same plant. Hard
seeds of the FI tended to give more hard-seeded plants in F z than
the progeny of soft seeds from FI plants. Freeman carried the
study through F 4 . He explained his results on the basis of two
pairs of factors, the heterozygous condition being intermediate
in soft-starch production. Since the endosperm results from
the union of two polar nuclei with a male generative cell, there
could be a range from zero to six factors for soft starch. The
type of soft starch worked with by Freeman is different from the
134 METHODS OF PLANT BREEDING
type called yellow berry, which is conditioned by inheritance but
is easily modified by environmental conditions.
Spike Density. Crosses between Triticum compactwn with
T. vulgare by Spillman (1909) and Gaines (1917) have shown one
main factor for compactness of head. In similar crosses, Parker
(1914) concluded that multiple factors were involved. Nilsson-
Ehle (191 la) studied crosses of Swedish Club (compact) with
Squarehead, a mid-dense-headed type, obtaining compact heads
in FI and segregation into compact, mid-dense, and lax in
FV He explained his results by the hypothesis of (7, a factor for
compactness epistatic to Li and L^ factors for length of internode
carried by Swedish Club and the recessive condition carried by
Squarehead, cc l\l\ hh- Fz plants of the phenotype c LiL% were
lax-headed. It is common, in crosses of vulgare with durum, to
obtain Emmer-like wheats with very dense heads. Stewart's
(1926) results from a cross of Sevicr with Federation, two varieties
of T. vulgare show that transgressive segregation for head density
may occur. Scvier is somewhat more dense than Federation.
The nature of segregation in F% was determined from F 3 progeny
trials of F% plants selected at random. Homozygous dense,
heterozygous, arid homozygous lax forms occurred in a 1:2:1
ratio, although the dense forms were more dense than Sevier and
the lax forms more lax than Federation.
Spring vs. Winter Habit. The main character that differenti-
ates spring from winter habit is heading behavior when wheat is
sown in the spring. In the spring-wheat areas of the United
States and Canada, winter wheat, when sown in the spring,
remains in the rosette stage and fails to head. Spring wheat
may be sown in the fall, and varieties of spring wheat often are
fall-sown in those sections where the winters are mild. Spring-
wheat varieties, as a rule, are less winter hardy than true winter
wheats.
In crosses between spring and winter wheat* the spring habit,
as a general rule, is completely dominant in FI, and segregation
occurs in F 2 . The type of ^2 ratio obtained without doubt
depends to a considerable extent upon the environmental con-
ditions used to differentiate spring from winter habit. Ratios
reported include simple ratios of spring to winter of 3 :1 by Cooper
(1923) and 15:1 by Nilsson-Leissner (1925); Vavilov and Kouz-
netsov (1921) and Aamodt (1923) obtained much more complex
INHERITANCE IN WHEAT 135
ratios. In the F% of Kanred X Marquis, Aamodt classified
plants from spring seeding for date of heading at weekly intervals
into eight weekly periods and into a winter group consisting of
types that failed to head. From 5253 F 2 plants, 980 headed as
early as the spring parent, and 442 were classified as winter. The
numbers of plants in other weekly periods for heading date were,
respectively, from early to late, 1503, 883, 568, 417, 313, 128, and
19. Plants heading in F 2 as early as Marquis bred true for spring
habit. Intermediates for date of heading bred true also in some
cases. Studies by Hayes and Aamodt (1927) of cold resistance
in crosses between Marquis with Minturki and Minhardi winter
wheats included a study of growth habit also. A late heading
type, when spring-sown, was selected that, when sown as winter
wheat, was rather highly winter-hardy. When recrossed with
Marquis and studied for cold resistance and for date of heading,
when spring-sown, there appeared to be almost complete correla-
tion between cold resistance and late heading (unpublished).
In general, there is a close correlation between winter habit and
cold resistance, but some wheats of winter habit are lacking in
high cold resistance. Some varieties of spring wheats have
considerably more resistance to winter killing when fall-sown
than other varieties.
Powers (1934) studied spring vs. winter habit of growth in a
cross of Hybrid 128 X Velvet Node. Under conditions at
Pullman, Washington, the parents and hybrids were classified
for date of ripening into weekly groups. The results were
explained by the interaction of three main factor pairs, where
A A, BB, and cc were factors for spring habit of growth and their
alleles for winter habit. A A was epistatic to bb and CC, BB to
aa and CC, and cc to aa and bb.
Stem-rust Reaction. There are 177 physiologic races of
Puccinia graminis tritici that have been differentiated by their
mode of reaction on 12 host varieties when inoculated with stem
rust in the greenhouse in the seedling stages [Stakman et al.
(1935), Johnson and Newton (1940), and Dickson (1939)]. It
is generally accepted that a variety of wheat, when resistant in
the seedling stage to a particular physiologic race of the dis-
ease organism, usually is resistant under field conditions from
heading to maturity to the same physiologic race. A variety
of wheat, however, may be highly resistant to one physiologic
136 METHODS OF PLANT BREEDING
race and completely susceptible to another. New physiologic
races originate (Craigie 1940) from hybridization on the bar-
berry, the alternate host of black-stem rust, and in the presence
of barberry bushes there is always the possibility of new physio-
logic races being developed.
The literature on the mode of inheritance of seedling reaction
is very extensive. Illustrations will be given of several types of
segregation. In a cross of H44-24 X Marquis (Goulden, Neatby,
and Welsh 1928), where H44 was resistant to physiologic race
36 and Marquis susceptible, results were explained by supposing
H44 to carry two duplicate factors for resistance, either alone in
the homozygous dominant condition leading to semiresistance.
Thus, the parental genotype of H44 was RiRi R^R^ arid of
Marquis T\TI r 2 r 2 . F% genotypes were as expected from a dihybrid
ratio with the genotypes RiRi R^R^ RiRi Rtfi, Rtfi R^R^ and
showing the resistant type Of phenotypic behavior,
R\Ri r 2 r 2 , rvi R^r^ and Rir^ r 2 r 2 being phenotypically
semiresistant, whereas the double recessive nr\ r 2 r 2 was highly
susceptible.
Harrington and Aamodt (1923) studied crosses between two
durum wheats; Pentad, resistant in the seedling stages to
physiologic race 34 and susceptible to race 1, and Miiidurn,
which reacts in a reciprocal way to these two races. A single
main genetic factor difference was responsible for reaction to each,
and the two factors were independently inherited.
In a cross of Kanred, immune to over 11 physiologic races,
with Marquis, which was susceptible to these same races (Aamodt
1923), immunity was dominant over resistance, and the manner
of reaction to all races to which Kanred was immune and Marquis
was susceptible was conditioned by a single genetic factor pair.
The few studies of seedling resistance that have been reviewed
briefly are representative of the many extensive studies of the
manner of inheritance of seedling reactions. The problem of
obtaining in a single variety resistance in the seedling stages to
all available races, and those races that may be found after further
study, has seemed rather difficult, since new physiologic races
are being found almost constantly, and the total number of
races is increasing rapidly from year to year. Recently, however,
the problem has appeared to be somewhat less difficult by the
discovery of several new wheats, notably wheats from the
INHERITANCE IN WHEAT
137
R.10
R55
R.14
R36
R.56
R.15
R38
R59
R.17
R.19
R.48
R.71
R.Z1
RHP
R.79
R.23
R.55
FIG. 17. Reaction in the seedling stages of a selection from the cross, Kenya
X Gular. Resistance to many physiologic races, both in the seedling and
mature plant stages, is due to a single dominant factor. (Courtesy of S. L.
Macindoe.)
138 METHODS OF PLANT BREEDING
Kenya Colony in Africa that have been described by Macindoe
(1931) as resistant to the prevalent races in Australia. Some of
these wheats have proved resistant in the seedling stages to %0
representative physiologic races (Peterson, Johnson, and Newton
1940) and have remained resistant also under field conditions,
where 30 prevalent Canadian races have been used to produce
the field epidemic.
The breeding of stem-rust-resistant vulgare types of wheat
originally consisted of attempts to transfer stem-rust resistance
from members of the Emmer wheat group to vulgare wheats by
crosses between the 14 and 21 chromosome species. Hayes,
Parker, and Kurtz weil (1920) obtained a vulgare type of wheat
resistant to stem rust, later named Marquillo, from a cross of
lurnillo, a durum variety, with Marquis. Marquillo proved less
resistant than lumillo, although under field conditions it has
continued to be moderately resistant to*a collection of prevalent
physiologic races. A sister selection of Marquillo was crossed
with a spring-wheat selection obtained from Kanred X Marquis
that carried the Kanred type of immunity to several races. From
this latter cross, Hayes, Stakman, and Aamodt (1925) concluded
that resistance in the stages from heading to maturity to a col-
lection of prevalent races was conditioned by two complementary
factors and that susceptibility was dominant to resistance. It
was found also that Marquillo and Thatcher, the latter selected
from the cross (Marquis X lumillo) X (Marquis X Kanred),
were highly susceptible in the seedling stages to several of the
prevalent races used in producing the field epidemic. The
resistance of the Marquillo type, conditioned by two main
factors in the field, proved independent in inheritance of the
Kanred near immunity to certain physiologic races.
A more satisfactory type of stem-rust resistance was obtained
by McFadden (1930) from crosses of a variety of Triticum
dicoccunij Yaraslov Ernmer, with Marquis. Two varieties
obtained from this cross, Hope and H44, although not entirely
satisfactory in agronomic characters, have in recent years been
used by practically all breeders as a source of stem-rust resistance.
These two wheats, like Thatcher and Marquillo, are susceptible
in the seedling stages to several physiologic races that occur
naturally both in United States and Canada, but both Hope and
H44 have proved highly resistant in the mature-plant stages in
INHERITANCE IN WHEAT 139
the field, from heading to maturity, to natural and artificial
epidemics of black-stem rust. As soon as McFadden obtained
these new wheats and before they were named, he generously
supplied seed to all breeders interested. It was soon evident,
as published by several workers at about the same time (Clark
and Ausemus 1928, Goulden, Neatby, and Welsh 1928) that the
type of resistance carried by Hope and H44 was simply inherited
in crosses with susceptible varieties of vulgare. Resistance is
dominant in Fi, and segregation in F% and later generations has
been found to be dependent upon one or two pairs of factors.
From crosses studied by Pan (1940) it seems probable that
resistant lines obtained from crosses with H44 carry the same
factors for resistance as Hope. In studies of ^3 lines from Hope
and H44 crosses with stem-rust-susceptible varieties of vulgare,
however, numerous workers have found a rather wide variation
in types of segregation including ratios of resistant to susceptible
of 9:7, 3:1, 15:1, 1:3, and 1:15 (Ausemus 1934, Churchward 1931,
1932). The important fact for the breeder is that resistant lines
continue to breed true for resistance in later generations.
Bunt Resistance. Farrar, in Australia, as early as 1901,
reported studies in the breeding of wheat varieties resistant to
bunt (Tilletia tritici (Bjerk.) Wint. and T. levis Kuhn). Gaines,
in Washington, has made extensive studies of inheritance of
bunt reaction. He classified his material as immune, resistant,
intermediate, and susceptible. In crosses of resistant and sus-
ceptible varieties, susceptibility was dominant, but when
immune varieties were used as one parent, there was a dominance
of immunity in FI. Although Gaines was unable to place his
results on a simple factorial basis, he found it possible to select
homozygous bunt-immune and bunt-resistant lines.
Briggs, working with nearly immune types of bunt resistance,
determined the genetic constitution of 10 bunt-resistant varieties.
The Martin factor is completely dominant, whereas the Turkey
and Hussar factors, when heterozygous, give an intermediate
reaction (Briggs 1933). There is some evidence of modifying
factors and Churchward (1931, 1932) has reported that the bunt
resistance of Florence is due to a single recessive factor. Recently
Briggs (1940) has concluded that there is a linkage between the
Martin and Turkey factors with a recombination value of 34,22
per cent,
140 METHODS OF PLANT BREEDING
TABLE 11. THE GENETIC CONSTITUTION OP 10 BUNT-KESISTANT VARIETIES
OF WHEAT (AFTER BRIGGS 1934)
Variety Bunt-resistant Factors
Martin MM hh it
White Odessa MMhhtt
Banner Berkeley .... MM hh it
Odessa MM hh tt
Sherman MM hh tt
Hussar ... MM HH tt
Selections 1418 and 1403 . . ... mm HH tt
Turkey 1558 . . mm hh TT
Turkey 3055 . . .... mm hh TT
Oro mmhh TT
Other Problems of Disease-resistance. Considerable infor-
mation is available regarding inheritance of resistance to other
diseases and insect pests, including reaction to scab, Hdmintho-
sporium sp., black chaff, mildew, leaf rust, stripe rust, and
Hessian fly. Varieties differ widely in their mode of reaction,
and for most of these pests it seems feasible to breed resistant
varieties. In many cases, however, sufficient information is not
available to place results on a genetic-factor basis.
Quantitative Characters. It seems reasonable to conclude
that all characters of crop plants are conditioned by genetic
factors. Yield of grain is a complex character that results from
the inheritance of genetic factors and their interaction under
particular conditions of environment. What is inherited is
manner of reaction under particular conditions and not the
character itself. Yield of grain is the end result of vigor of
plant, as expressed in number of heads, number of kernels per
spike and spikelet, and size of individual kernel. Anything that
interferes with the normal development of the plant, including
injury from diseases and unfavorable environmental conditions,
affects yield. The usual method adopted by the breeder, when
quantitative characters are concerned, is to select parents of
good yielding ability with desirable characters, including those
particular qualities for which the crop is used, select during the
segregating generations for the characters desired, and test
hybrids for yielding ability and other characters before deciding
which is the more desirable,
CHAPTER X
INHERITANCE IN OATS
Cytological studies made by Kihara, Nishiyama, and others
place A vena species in three groups, based on differences in chro-
mosome numbers. These have been summarized by Stanton
(1936):
Group 1. n = 7 chromosomes. Avena brevis Roth (short oat); A.
wicstii Steudel (desert oat); A. strigosa Schreb. (sand oat) and A. nudi-
brevis Vav. (small seeded naked oat).
Group 2. n = 14 chromosomes. A vena barbata Pott (slender oat)
and A. abyssinica Hochst. (Abyssinian oat).
Groups, n 21 chromosomes. Avenafatua"L. (common wild oat);
A. saliva L., including A. oricntalis Schreb. (common white or northern
oat); A. nuda L. (hull-less oat); A. sterilis L. including A. ludoviciana
Dur. (wild red or animated oat); and A. byzantina C. Koch, including
A. sterilis algcricnsis Trabut (cultivated red oat).
Kihara and Nishiyama (1932) have reported extensive studies
of species crosses. Interspecific crosses between species belong-
ing to the same chromosome group can be made easily. Crosses
between the 14 X 28 chromosome species are relatively easy but
produced fertile seeds only when the 28-ehromosome species was
the female. Crosses between the 14- and 42-chromosome species
are difficult, and only a few successful crosses have been obtained.
Viable seeds were produced only when the n = 21 species was the
female. Reciprocal crosses between the 28- and 42-chromosome
species gave well-developed kernels that germinated well.
Chromosome affinities in species crosses have not been com-
pletely worked out. A table by Nishiyama (1929) summarized
some relationships by listing the number of bivalent associations
inclusive of trivalents in hybrids between different species. He
says, "If two parents differ in chromosome numbers, as many
bivalents as the lower chromosome number of one parent may
be expected. A full affinity between two species is, therefore,
represented as LOGO and no affinity as 0.000,"
141
142 METHODS OF PLANT BREEDING
These results indicate that A. barbata is not closely related to
A. fatua. A. strigosa (n = 7) when crossed with A. barbata
(n 14) showed more than seven bivalents, due possibly to
autosyndesis of barbata chromosomes.
TABLE 12. CHROMOSOME AFFINITIES IN SPECIES CROSSES OF Avena
Avena strigosa Avena saliva 0.983 Avena byzantina
i f
0.998 0.986
iwjd at/i frj,
1.041
Avena barbata 0.456 Avena fatua 0.992 Avena sterilis
0.675
As would be expected, crosses between species with different
chromosome numbers are partly or nearly completely sterile
because of irregularities of chromosome behavior during meiosis
in the FI hybrid.
In the first-division metaphase of crosses between species that
differ in chromosome numbers, bivalents and trivalents form a
normal equatorial plate with the univalents scattered throughout
the cell. Univalents divide equationally and pass to the poles,
being included in the daughter nuclei, except for a few lagging
chromosomes. The second division is irregular, since the uni-
valents that have already divided equationally pass to the poles
at random. There are many lagging chromosomes. In a cross
between A. barbata with A. strigosa, 7 bivalents inclusive of
trivalents are found commonly, and in some cases 8 or 9 bivalents.
Trivalents are frequent. In crosses between A. barbata with
A. fatua, the number of bivalents varied from two to eleven,
with 1 to 4 trivalents; in the FI crosses of A. barbata X A.
sterilis, 7 to 13 bivalents were found, inclusive of to 4 trivalents.
These results are not widely different from those in species crosses
in wheat and give some reason for the belief that desirable charac-
ters from species with lower chromosome numbers can be trans-
ferred to the cultivated species with 42 chromosomes.
Cultivated varieties of oats belong chiefly to the species A.
saliva j including the side-oat group A, sativa orientalis, commonly
believed to have been derived from the wild oat (A. fatua L,)
and the red oat varieties of A. 'byzantina, derived from the wild
red oat, A. sterilis. Crosses between different species belonging
to the 42-chromoeome group are highly fertile, although there is
INHERITANCE IN OATS 143
some evidence of abnormal chromosome behavior possible
because of structural changes in one or more chromosomes within
the various sets. This may cause trivalent, quadrivalent, or
lagging chromosomes at meiosis. Nishiyama (1929) concludes,
"All hexaploid hybrids have normal bivalents in the majority of
P.M.C. at the metaphase of the first division. Sometimes l*-4
univalents and certain chromosome complexes are found together
with normal bivalents. These irregularities are probably caused
by mating between semihomologous chromosomes, not being
normal partners. 7 '
Inheritance of Characters in Crosses between 42-chromosome
Species. Surface (1916), Philp (1933), and others have studied
the inheritance of characters in crosses between Avena fatua
and A. saliva. Characters associated with the fatua b^tse on the
grain of the lower floret that are completely correlated in inherit-
ance include (1) heavy awn on the lower grain, (2) awn on the
upper grain, (3) fatua base on the upper grain, (4) pubescence
on the rachilla of the lower and upper grain, (5) pubescence on
all sides of the lower grain and on the base of the upper grain.
Philp explained these results by a factor C carried by the fatua
parent that was partially dominant to c carried by A. sativa.
It was suggested by Philp that the chromosomes carrying C
in A. fatua and c in A. sativa were not entirely homologous and
that the factor pair Cc responsible for a group of linked charac-
ters was inherited as a group complex. A partial lack of chromo-
some homologies was given as the probable reason for the
complete linkage of the several character pairs.
The upper grains of the floret are persistent to their rachillas
in A. sterilis and A. byzantina, which differentiates them from
A. fatua and A. sativa , whereas cultivated varieties of A. byzan-
Una differ from varieties of A, sativa in that there is a well-defined
deep, oval cavity or "sucker mouth" on the base of the lemma of
A. byzantina. These differences were illustrated in Chap. II.
Different investigators, including Fraser (1919), Hayes, Moore,
and Stakman (1939), and Torrie (1939), have studied linkage
relations of differential characters in crosses between varieties of
A. sativa with A. byzantina. Coffman, Parker, and Quisenberry
(1925) studied variability in Burt oats, belonging to A. byzantina }
with particular reference to the following characters, using three
characters in their classification:
144 , * METHODS OF PLANT BREEDING
Spikekt disarticulation, or the separation of the lowet floret of
the oat spikelet from the axis of the spikelet, was divided into
three groups: (1) abscission, leaving a well-defined cavity in the
face of the callus on the base of the lemma of the lower grain,
(2) disarticulation by fracture, resulting in a rough fractured
surface with little or no cavity in the base of the lemma, charac-
teristic of A. saliva j (3) disarticulation by semiabscission, more
or less intermediate between 1 and 2. Groups 1 and 3 charac-
terize varieties of A. byzantina and homozygous segregates of the
byzantina type from A. saliva X A. byzantina.
Floret disjunction, or the separation of the second or upper floret
from the lower, was also classified in three groups: (1) disjunction
by basifracture, the rachilla segment breaking near its base and
remaining firmly attached to the upper floret, (2) disjunction by
disarticulation at the apex of the rachilla segment, the rachilla
segment remaining attached to the lower floret, the normal
method in A. sativa, and (3) disjunction by heterofracture, the
break occurring more or less intermediate between (1) and (2).
Basal hairs refer to conspicuous bristles on the base of the lower
floret. Three classifications are given: (1) abundant long, (2)
abundant mid-length, and (3) few.
In a series of crosses between Bond, A. byzantina, and cultivated
varieties of A. saliva, several workers have found linkages for
various character pairs. The linkages for the following charac-
ters were given by Hayes, Moore, and Stakman (1939).
1. Spikelet disarticulation and basal hair development, recom-
bination value 2.7 per cent.
2. Floret disjunction (one of two genes involved) and basal
hair development, recombination value 24.0 per cent.
3. Spikelet disarticulation and floret disjunction, recombina-
tion value 25.7 per cent.
Torrie (1939) observed linkage relations, in crosses between A.
sativa, Iowa 444, and A. byzantina, Bond, for character differ-
ences including spikelet disarticulation, floret disjunction,
rachilla attachment, basal hair length, awning, and red lemma
color. The exact linear order of the genes was not accurately
determined. The results indicated, however, that it was possible
to obtain new combinations of these characters if desired.
Differences IE Awn Development. Varieties of oats differ
widely in awn development, both in the number of awns on the
INHERITANCE IN OATS 145
upper and lower florets and in the degree of development of awns.
The extent of awn development in a pure line varies rather widely
from plant to plant and from one panicle to another on the same
plant because of environmental conditions. Under uniform
conditions, pure lines may be selected that show a range in awn
development from nearly awnless to lines that have strongly
developed awns on both the upper and lower florets. In some
crosses, 3:1 or 1:2:1 ratios have been obtained when the separa-
tion is made into the larger groups, although, as a rule, minor
modifying factors that modify the degree of development of the
main factor pair are also involved.
Nilsson Ehle (19116) and Love and Craig (19186) found evi-
dence that the gene for yellow lemma color inhibited the develop-
ment of awns. Fraser (1919) studied a cross between Sixty-day,
with yellow grains and no awns, with Burt, A. byzantinaj with
weak awns on the lower floret and frequently on the upper.
In F 2 , there was a ratio of fully awned (like Burt) to awnless of
1:3. The fully awned plants bred true in jP 3 . The degree of
development of the awns ranged from weak awns like Burt to
strong awns that were stiff and long, the strong awn being sharply
twisted at the base, with a sharp bend about three-eighths of the
way from the base to the tip. In crosses between logold, with
weak to intermediate development of awns and with less than
50 per cent of the lower florets bearing an awn, and Bond, with
100 per cent weak awns in the lower florets, a range in F% from
strong to weak awns was obtained and also a range from 25 per
cent awned to fully awned (Hayes et al. 1939).
Color of Grain. The color of the lemma has been classified as
black, brownish red, gray, yellow, and white. Intensity of color
is influenced by environmental conditions, and it is sometimes
difficult to differentiate yellow and white. With bright sunshine
during the later stages of development, the intensity of color is
deeper than when wet, cloudy weather conditions prevail.
Black is epistatic to gray and yellow (Nilsson-Ehle 1909,
Surface 1916), (Love and Craig 19186), and gray is epistatic to
yellow. Black vs. colorless, gray vs. colorless, and yellow vs.
colorless segregated on a single factor basis in some crosses. In
other crosses, there may be duplicate factors for black and for
yellow. In a cross of Sixty-day, which produces yellow grain,
with Burt, which produces brownish yellow, Fraser (1919)
146
METHODS OF PLANT BREEDING
For descriptive legend see opposite page.
INHERITANCE IN OATS 147
obtained a ratio of 48 red: 15 yellow: 1 white4n F*. Apparently
Burt carries a factor for red, R, and for yellow, F, red being
epistatic to yellow. The factor for yellow in Sixty-day is inde-
pendent in inheritance of the factor for yellow carried by Burt.
In crosses of Bond, reddish yellow X Iowa 444, colorless, Torrie
(1939) concluded that the Bond parent carried two dominant
factors, one for reddish color and one for yellow, that were inde-
pendently inherited. Philp (1933) concluded that black and
gray were independently inherited.
Hulled vs. Hull-less. The Avena nuda species has been dif-
ferentiated on the basis of its hull-less condition, Love and
McRostie (1919), in crosses of hulled X hull-less, obtained an
intermediate condition in Fi and a ratio in F% of l:2:.l. Some
evidence was given of a factor that modified the percentage of
hulled grains on heterozygous plants. Philp (1933) obtained
some hull-less F% plants in crosses of A. saliva X A. fatua,
although they were not completely hull-less. He reports A.
nuda plants from crosses of the A. sativa varieties made by
W. Bobb. The results can be explained by supposing that A.
sativa carries two types of chromosome complexes, called Z and
Zj with Z epistatic to z. The Z complex carries a factor for hulled
while z carries a factor for naked. A change of pairing whereby
Z occasionally pairs with z will lead to the production of hull-less
plants.
Spreading vs. Side Panicle. Nilsson-Ehle explained a cross
between spreading vs. side-panicle varieties on the basis of
duplicate factors, either factor in the dominant condition pro-
ducing an open panicle. Gaines (1917) and Garber (1922) found
it difficult to separate spreading and side-panicle forms in
segregating generations. Either 3:1 or 15:1 ratios would be
expected in later generations if duplicate factors were involved
from crosses of spreading vs. side-panicled varieties. There
FIG. 18. Panicle arid floral structure of oats,
1. Branch of oat panicle.
2. Spikeiet, showing tertiary floret just after blooming: (a) primary floret.
3. Spikeiet, showing floral parts: (a) outer glume; (b) flowering glume; (c) palea;
(d) lodicules; (e) anther; (/) stigma; (0) secondary floret; (h) awn.
4. Outer parts removed, showing sexual organs.
6. Longitudinal section of ovary.
6. Anther.
7. Showing outer and flowering glume of lower spikelet removed: (a) lodicules,
and sexual organs.
Size; 1, 2, about X; 3, about 2x; 4, 5, 6 fereatly enlarged; 7, about 2X.
148 METHODS OF PLANT BREEDING
probably are modifying factors that make the separation between
open and side panicle difficult in some crosses.
Pubescence. Cultivated varieties of sativa oats differ in the
amount and in the presence of basal hairs on each side of the
callus of the lower floret. One or two pairs of factors are involved
in various crosses. Transgressive segregation, therefore, occurs
in some crosses, forms being obtained in F 2 that are more pubes-
cent than either parent or that lack pubescence. Pubescence
on the back of the lower grain, the wild type of A.fatua, is domi-
nant to the glabrous condition and may be controlled by one or
two duplicate factors. One of these is closely linked and in some
crosses completely linked with a factor for black grain color
(Nilsson-Ehle 1909, Surface 1916, Love and Craig 19186, and
Philp 1933).
DISEASE REACTIONS
Three important diseases of oats are stem rust, Puccinia
graminis avenae Eriks. & Henn., crown rust, P. coronata Corda,
and the smuts, Ustilago avenae (Pers.) Jens, and U. levis (K. & S.)
Magn.
Physiological specialization occurs for all three diseases. Dick-
son (1939) listed 9 physiologic races of stem rust and 44 of
crown rust; Reed (1940) listed 29 of U. avenae and 14 of U. lens.
In a breeding program, it is essential to use physiologic races
prevalent in the locality to produce the artificial disease epidemic.
Stem Rust. Varieties of oats are available that are resistant
to several of the races common in the sections where stem rust
frequently causes severe injury to susceptible varieties. logold
and Rainbow are resistant to races 1, 2, 3, 5, and 7, and White
Russian and derivatives are resistant to races 1, 2, 5, 8, and 9.
Resistance to the five races 1, 2, 3, 5, and 7, to the three races
1, 2, and 5 and probably 8 and 9, and susceptibility to all five
races form an allelic series (Smith 1934), and in any one cross the
only homozygous types that can be obtained are the parental
types. The resistance of logold and Rainbow under both field
and greenhouse conditions to the races to which logold, Rain-
bow, and White Russian are resistant is of somewhat higher
type than that of White Russian.
Welsh (1931) pointed out that resistance of Hajira to races 1,
2, 3, 5, and 7 was governed by the same factor pair. Joanette
INHERITANCE IN OATS 149
is resistant to race 4, and in crosses with Hajira, segregation for
rust reaction to race 4 was on the basis of 9 resistant: 7 suscepti-
ble. From a test of 21 lines breeding true for resistance to race
4, about half of these were resistant also to races 1, 2, 3, 5, arid 7.
Welsh (1937) has reported obtaining strains resistant to race 6
from crosses of Hajira with Joanette and explains the results on
the basis of transgressive segregation. It is of some interest
FIQ. 19. Culms of resistant and susceptible varieties of oats. From left
to right: Victory, susceptible to stem rust; a susceptible Ft plant of Victory X
White Russian; a resistant Fz plant of Victory X White Russian f resistant
White Russian.
that White Russian was semiresistant to race 6 under field con-
ditions in trials made by Welsh.
Although resistance to all races is to be desired, resistance of
Rusota and Rainbow to races 1, 2, 3, 5, and 7 and of White
Russian, Anthony, and Minrus to races 1, 2, 5, and perhaps 8
and 9 has protected these varieties from serious injury from stem
rust under Minnesota conditions for many years when suscepti-
ble varieties such as Victory are often severely injured.
Crown Rust. Although some -varieties have been available
that in some seasons show resistance to crown rust, the intro-
150 METHODS OF PLANT BREEDING
ductioii of Victoria from South America and Bond from Australia
(Stanton & Murphy 1933) has furnished a basis for the breeding
of resistant varieties, since both Bond and Victoria are resistant
to many races of crown rust.
Using Victoria as one parent in crosses with susceptible varie-
ties, Smith (1934) concluded that resistance was a partial
dominant in Fi. Variable infection made it impossible to decide
the number of factors involved. Stanton (1936) indicates a
single factor pair with resistance dominant.
In crosses of Bond with susceptible varieties, Hayes et al.
(1939) concluded that two pairs of factors were involved and
placed segregation on a 9 : 7 basis, whereas Torrie (1939) explained
crown-rust inheritance in crosses of Bond with Iowa 444 on the
basis of two pairs of factors, Ss for resistance vs. susceptibility
and lij a factor pair that in the dominant condition partly
inhibited the effect of S.
Correlated Inheritance of Reaction to Three Diseases.
Stanton et al. (1934) obtained selections from crosses of Victoria X
Richland that were resistant to the three diseases, and Murphy,
Stanton, and Coffman (1936) reported selections from crosses
where Bond was used as one parent that were resistant also to
the three diseases. Hayes et al. (1939) and Torrie (1939) in
crosses of Bond with varieties of A vena saliva found that reaction
to all three diseases was inherited independently and found no
evidence of association of reaction to the three diseases and other
characters differentiating A. byzantina and A. saliva.
Smuts. Reed (1940) summarized the reaction of a consider-
able group of varieties and species of oats to all physiologic
races of both species of smuts. Markton, a well-known variety,
Navarro (Stanton 1933), and Victoria are resistant to all races
of both smuts. Black Mesdag has been used extensively in
crosses and is resistant to all races of Ustilago avenae and resistant
to 10 of the 14 races of U. levis. So far as tested, A. barbata
is susceptible to all races of both smuts. A few varieties of oats
are susceptible to most races. Canadian, for example, is sus-
ceptible to 28 races of loose smut and 13 of covered, being
resistant to one race of each.
In crosses of Monarch selection X Black Mesdag (Stanton,
Reed, & Coffman 1934), inoculated with races of 7. avenae from
Missouri, resistance was a dominant, and segregation on a 3:1
INHERITANCE IN OATS 151
basis occurred. Cresses of Markton X Black Mesdag inoculated
with U. avenae, where both parents were resistant, gave some
susceptible F$ progenies.
Hayes et al. (1928) studied crosses of Black Mesdag with other
A. saliva selections, using a mixture of smuts for inoculation, and
obtained results indicating that the Black Mesdag resistance to
both smuts was due to the action of two pairs of factors, R for
high resistance and I for immunity carried by Black Mesdag. A
selection from this cross that has proved highly resistant to a mix-
ture of races of smuts at University Farm, St. Paul, Minnesbta,
was crossed with Bond, resistant also to the races used. Some
susceptible plants and lines occurred in F% and F 3 , respectively.
In crosses of Bond with susceptible varieties, resistance was
dominant and the segregation was on a 3:1 basis. Results of
this nature are common in polyploids of an amphidiploid nature
where cases of duplicate or triplicate factors that condition the
development of a character are of relatively frequent occurrence.
QUANTITATIVE CHARACTERS
Many characters of oats of interest to the breeder are undoubt-
edly due to the interaction of multiple factors. These include
such characters as data of maturity, height of plant, resistance
to lodging, number of culms, winter hardiness, drought resist-
ance, percentage of hull, weight per bushel, and yielding ability.
It is important for the breeder to analyze the varieties that are
used as parents for all characters of importance and select during
the segregating generations, under controlled conditions when
possible, for the characters desired.
CHAPTER XI
INHERITANCE IN BARLEY
CLASSIFICATION AND GENETICS OF BARLEY SPECIES
Harlan (1918) classified barley into four species, essentially on
the basis of fertility of the lateral spikelets. The following key
is taken from Harlan's paper:
All spikelets fertile (6-rowed barley) :
Lemmas of all flowers awn eel or hooded Hordeum vulgar -e L.
Lemmas of lateral flowers bearing neither awns nor hoods
//, intermedium Keke.
Only the central spikelets fertile (2-rowed barley) :
Lateral spikelets consisting of outer glumes, lemma, palea, raehilla, arid
usually rudiments of the sexual organs Hordeum distichon L.
Lateral spikelets reduced, usually to only the outer glumes and raehilla,
rarely more than one flowering glume present, and never rudiments of
sexual organs Hordeum defidens Steud.
The H '. intermedium group would be classified more accurately
by the statement: Central spikelets fertile, lateral spikelets
partially fertile.
A single-factor difference for type of head is found in crosses of
some varieties of //. vulgare X H. defidens; H. vulgare X H.
distichon; and //. distichon X H. defidens. Engledow (1924)
and Hor (1924) concluded that an allelic series of factors differ-
entiated the type of lateral florets found in these three species.
In some crosses of varieties of //. vulgare with H. distichon,
as has been already mentioned, a segregation of two-rowed
(VV) : intermediate (Vv) : six-rowed (vv) of 1:2:1 is obtained. In
cases of monohybrid segregation, the lateral florets are usually
infertile but will always be awn-pointed. In other crosses of
H. vulgare with H. distichon } seven classes may be differentiated
by the breeding behavior in F 3 . Harlan and Hayes (1920) gave
the first complete genetic analysis of the results from such
crosses, explaining the results on the basis of two factor pairs.
Robertson (1933) obtained similar results. Both obtained true-
152
INHERITANCE IN BARLEY 153
breeding intermedium types in F 3 . The lateral spikelets in the
intermedium obtained by Harlan and Hayes (1920) were par-
tially fertile, varying from 18 to 55 per cent in different F s lines.
In the intermedium obtained by Robertson (1933), the lateral
spikelets were infertile, i.e., less than 2 per cent fertile. Leonard
FIG. 20. Heads of the cultivated species of barley. From left to right,
Hordeum vulgare, H. intermedium (fertile), H. intermedium (infertile), H, dis-
tichon, H. deficiens.
(1940) found that the fertile, infertile, and nonintermedium types
were differentiated by genes belonging to a multiple-allelic series,
designated as I h l h , //, and ii, respectively.
Intermedium barley can be classified on the basis of the
rounded lemmas of the lateral florets, which are never awn-
pointed. This condition is expressed only in the presence of VV.
In the presence of Vv, the lateral florets are always awn-pointed.
154
METHODS OF PLANT BREEDING
These are designated intermediates. Varieties that are geno-
typically vv are six-rowed, with complete fertility of the lateral
florets. These may be vv I h l h j vv II, or vv ii.
The genotype of an unknown six-rowed variety for the inter-
medium series may be determined by crossing it with a tester
strain of known genotype, such as Nigrinudum, which is known
to be VV II. The term infertile intermedium may be used when
less than 2 per cent of the lateral spikelets are fertile and the term
fertile intermedium used to designate those types with more than
2 per cent (usually 10 to 60 per cent) of fertile lateral florets.
The following scheme will illustrate how the genotypic constitu-
tion of the six-rowed variety may be determined when crossed
with a variety that has the genotype VV II. The phenotypes of
the Fi and Ft generations are given for crosses of VV II with three
different homozygous six-rowed genotypes.
Phenotype of:
Genotype of
six-rowed
variety
Fertile intermediate.
Fertile intermediate .
Infertile intermediate
Infertile intermedium, fertile in-
termediate, and 6-rowed in
ratio of 1:2:1
Infertile intermedium, fertile in-
termedium, fertile intermediate,
and 6-rowed in ratio of 3:1:8:4.
The distinguishing feature is the
presence of fertile intermediums.
2-rowed, infertile intermedium,
infertile intermediate, fertile in-
termediate, and 6-rowed in a
ratio of 3:1:6:2:4. The dis-
tinguishing feature is the pro-
duction of 2-rowed segregates
but no fertile intermedium
mil
vv I h l h
VV
It is sometimes difficult to Determine morphologically whether
barley varieties are genetically of the distichon type (VV ii) or
are infertile intermediums (VV II). To discriminate between
them, they may be crossed to tester strains of the genotype vv II.
If segregation is on a monohybrid basis, the genotype of the two-
rowed parent is VV /I, and the variety is an infertile inter-
INHERITANCE IN BARLEY^ 155
medium. If a dihybrid segregation occurs, the two-rowed
parent is a true two-rowed barley with the genotype VV ii.
The four species of barley described by Harlan (1918) are
differentiated genetically by only two factor pairs and their allelek
CHROMOSOME NUMBER IN GENUS HORDEUM
"),
In the genus Hordeum, as in Triticum $nd Avena, the basic
chromosome number is seven pairs. Multiples of this basic
number are obtained also. Numerous investigators have
reported the chromosome number of different species, and some
of these are listed below:
7 pairs of chromosomes:
Hordeum bulbosum, H. deficiens, H. distichon, H. gussoneanum, H. hexa-
stichum, H. intermedium, H. jubatum, H. murinum, H. nodosum, H.
pusillumj H. spontaneum, H. vulgare
14 pairs of chromosomes:
H. bulbosum, H. jubatum, H. murinum, H, secalinum
21 pairs of chromosomes:
H. nodosum
The species bulbosum, jubatum, and murinum have been
reported by different investigators as having either 7 or 14 pairs
of chromosomes. H. nodosum has been reported as having 7 or
21 pairs. The economic species all have 7 pairs of chromosomes.
LINKAGE GROUPS
There are numerous characters in barley that are easily dif-
ferentiated. Since the number of chromosome pairs is seven for
each of the four cultivated species, barley has been used extensively
in studies of linkage relations. More than one hundred different
characters have been investigated. Robertson, Wiebe, and
Immer (1941) summarized the known* linkage information and
suggested symbols to be used in designating the various charac-
ters. In Table 13 are given some of the characters that are
known to be simply inherited and that have been placed in one
of the linkage groups.
It is of some interest to note that to date the only four factor
pairs known for group 6 involve lethal seedlings. In all other
chromosomes, easily differentiated, completely viable characters
are available.
Internode Length in the Racbis of the Spike. Varieties of
barley vary greatly in density of the head, as measured by length
156 METHODS OF PLANT BBEEDIN6
TABLE 13. SIMPLY INHERITED CHARACTERS IN DIFFERENT LINKAGE GROUPS
Character Differences Symbol
Group 1:
Non-6-rowed vs. 6-rowed Vv
Red vs. white pericarp Re\re\
Purple vs. white lemma Pp
Purple vs. white straw Prpr
Toothed vs. untoothed lemma . Gg
Awnless vs. awned Lklk
Normal vs. albino seedlings Aa
Normal vs. albino seedlings At&i
Normal vs. albino seedlings A 404
Green vs. chlorina seedlings Ff
Green vs. virescent seedlings Yy
Green vs. orange seedlings Oror
Group 2:
Black vs. white lemma and pericarp Bb
Normal vs. "third outer glume" Trd trd
Normal vs. albino seedlings At i
Group 3:
Hulled vs. naked Nn
Normal vs. albino seedlings A cZ a cz
Dense vs. lax head LI
Group 4:
Hooded vs. awned Kk
Blue vs. white aieurone Bl bl
Fertile intermedium, infertile intermedium, and noninter-
medium /*, /, i
Group 5:
Rough vs. smooth-awned Rr
Long vs. short-haired rachilla , . , Ss
White vs. orange lemma Oo
Normal vs. albino seedlings Ab a&
Red vs. white pericarp Re re
Group 6:
Green vs. xantha seedlings * X e x e
Green vs, xantha seedlings X x
Green vs. albino seedlings A e a e
Green vs. albino seedlings A n a n
Group 7:
Normal vs. brachytic Br br
Green vs. chlorina seedlings. . . / F e f e
Green vs. virescent seedlings Y c y^
Resistance vs. susceptibility to Pucdnia graminis , , , Tt
INHERITANCE IN BARLEY 157
of the internodes of the spike. The density varies comparatively
little from year to year. Hayes and Harlan (1920) studied the
mode of inheritance of internode length in five crosses. In two
crosses, a single-factor-pair difference explained the results
satisfactorily. Short internode length was dominant in one of
these crosses, but the head density in the second cross was inter-
mediate in FI. In another cross, a broad difference of two factor
pairs was indicated by the segregation in F 2 and F 3 .
In the cross of Hanna X Zeocriton (see Table 14), lax and
dense varieties, respectively, the F 2 ranged from above the modal
FIG. 21. Average spikes of the Zeocriton (left), Hanna (right), and four
homozygous lines. Mean densities are as follows: Zeocriton, 1.9 mm.; Hanna X
Zeocriton, 448-1, 2.3 mm.; 448-5, 2.9 mm.; 448-11-3, 3.7 mm.; 448-16, 4.3 mm.;
Hanna, 4.6 mm. (After Hayes and Garber, 1927.)
class of Hanna to the modal class of Zeocriton, even though only
141 individuals were studied. F 3 families were grown from F%
plants representing different densities. Progenies from selected
plants in certain FZ lines were tested further in F 4 . Some F$
lines bred comparatively true, the range for density being no
greater than for the parental varieties. Other F$ lines were as
variable as the F 2 generation and still others more variable than
the parents but less variable than the F%. Typical heads of the
parents and segregates from homozygous lines are illustrated in
Fig. 21.
Homozygous lines differing in density were obtained in F% and
F 4 . The homozygotes appeared to fall in groups. The general
158
METHODS OF PLANT BREEDING
<flcNNlOCOOCOOvlOOiO<N
QO'^CO'O'^ | O>OiCO 1 ^COPOOO"5' <1 *00'^ | NO>
.
OOOQOrHOOOOO
OOOrHOOOOOOOOOOOOOO
4_l 4_| _^| +| -|_| +1 -|-| +| -}-| 4-1 +1 -f-| +| +| -f | +| -fl -H
o
iCO^OO^^^^awco
^o.o^^cooot-coco
Ot-iOrHCOOSt-^CO^^WNCO^lOCOr-J
ooooooooooo
oooooooooooooooooo
05
ooooooooooo
oooooooooooooooooo
Tf<Tj<^rH<NCO<*l<NC<J<NtN
(NMcocNWcvswT^eocoericoeoeO'*^^^
3
5*33*SSSS
W 22 r ^' :D S' Hl0lOI> ^' rH 1CO ' '^cO CO
9'S
r " H ' '!'.'*'
f'S
; ; "- ;
S'S
: . ^ . . .
O'S
^ 2 ' : '
. rH CO rH
a 8 >
N "" co ^ ' ' . ;
. M . . : : : ^ i .
* 9't
IO5 O O
<N CO
: ": ro :^.: : co : rHoo,.H^
' *'*
(N CO CO (N
rH CO CO
: : TO : 2 . : ' ~ l>% : S ^
1 ?*
2 co S : ' *" : : ' ' '
. ; ' w .* *" .' S ' ^ CNI S5 2 o *""
-3 o>
IO C5S CO CO
CO CO rH ...
rH . rH CO (N CO rH
& 8*S
<% > ~ . . ** . ' . ;
.. CN'OlrH^lO^CO^O.^^^^
| 9"
CO rH . . . OJ < - ;
_^CM -COCO : OOrH05COCO^ ;C4rH
ft F
. : :2 - w ^
." .' *"* . J^ ^ ." ?5 "* * N ^ 2 '
S &*e
I . : : :2 : . S
CO O r* .lOiO -JOO -(NOOOs -CO
Tfl rH rHCO-rHrH(N
| O'g
: : : : : ^ : : * ^ S
Tt< -CNOO -OOrH -COI> -COOSCO -rH
rH Th CN rH
C5
00 rH iO O rH
CO -t>-J^- O rH rH -COrHtO
gg 9'2
00 CO O CO M
rH - <N rH 10
CO CO rH CO O .... . .
5 rg
*. ; .' i ^ t>. <N
g ^ ^ ..:.:.
S'2
1. (N 00 00 O rH
. . CN rH 00 0* -
O rH
0*2
. . , ^4 b. ^H (N rH
N CO
8"!
. : : S S ^ ^ w
CO CO CO -
9*1
. . .1C rH
.<-? t3
O) 1> ^ O
00
d *o *
CO rH (N O> CN CO t*
OiC^COCOCNCOCO^COOOCOCOCOCO^'^CO''*
Q a
. CO ^ rH <M
TH
Sa
cOt-OOKOOOKOOKt-QO
^OOXl-OOOOOOQOt-OOOO^OOOOOOt-OOK
OJOlOsOiOSOOiCTSOSOSOs
0>0505C^050<^C^05Cft05C750i05a.350505
o5 S
8|
ft, ft, ft, ft, ft, fe, fe, *, fe, fc, fe,
&, fe, fe, P^ fc, fe, B, ^ r^ fe, &,fc, k B^ fe, k ft, fc,
.
0)
^
1
: : 5 : : 3
:::::::::::::::: |
5 :1 : :1
: ! '::::':'::: '"&
o ! Tt< ! ' ^
jssjjisl^^^^^ssj^co
i i siiin
^^li^l^^m^i^
INHERITANCE IN BARLEY
159
^
i
J
/
\
/
\
^
K
L
ocii
ton
\
^ aT
w
^
\
1X .
if i
2*19
'-Til
-N
191(1
S
""-
<
X-
-^
918
X
""X
/
*~^
V
t
1
N ^
/
\
/
\
!
\
S
/
X,
\
/
)
A
f
/;
>
\
/
448
I
fa
k\
X
i48-
l "i
448
ik.
X
""'s.
nature of the results is illustrated by Fig. 22. The results could
be explained on a genetic basis by the hypothesis that the parent
varieties differed by three independently inherited factors.
These factors were considered to have a cumulative effect. Other
factors, having smaller effects, doubtless were present also and
modified the expression of the main density factors.
Wexelsen (1934), from studies of six crosses involving five
different varieties, found that internode length in different crosses
50
40
80
20
10
3
. 50
I 40
5 20 i
1 10
6 o
30
20
10
l.*1.6 1.8 2,0 2.2 2.12.6 2.8 3-0 3-2 3.4 8.6 3-8 4.0 4*2 4,4 4.6 4.8 5.0 52 5.4> 6,6
Density in MM
FIG. 22. Diagrams showing the densities of parental forms and of the Fi
generation in a cross between the Zeocriton and Hanna barleys (upper), of four
pure lines (middle), and of several heterozygous lines (lower). (After Hayes
and Harlan, 1920.)
was differentiated by one to five factor pairs. A total of six
factor pairs appeared to be involved in these crosses. These
factors had different effects on internode length when hetero-
zygous, some heterozygous types being intermediate, one being
near the short, and another near the long internode parental
type. One of the internode-length-factor pairs (L^k) was found
to be linked with rough vs. smooth awns (Rr) and long vs. short-
haired rachilla (Ss), and another (L 4 ^) was linked with non-six
rowed vs. six-rowed head type (Vv).
Reaction to Helminthosporium sativum. The best proof that
quantitative characters are inherited in the same manner as
qualitative characters has been obtained from linkage studies.
Quantitative characters may be correlated with qualitative!
160 METHODS OF PLANT BREEDING
characters when the mode of inheritance and linkage relations
of the qualitative characters are known. By means of such
studies, it is possible frequently to determine the minimum
number of genes controlling the quantitative character in crosses
between known varieties. This mode of attack was used by
Griff ee (1925) in a rather extensive study of reaction to spot
blotch Helminthosporium sativum P. K. & B.
The contrasted characters of the parent varieties were as
follows :
Svanhals Lion
White hull and pericarp Black hull and pericarp
2-rowed (distichon) 6-rowed (vulgare)
Rough awn Smooth awn
Resistant to spot blotch Susceptible to spot blotch
Each of the character pairs, black vs. white, two-rowed vs.
six-rowed, and rough vs. smooth awn, are known to be dependent
upon single-factor differences and to be independently inherited.
By considering each of these character pairs separately, a definite
association was found in F% (each Ft plant was tested by growing
and examining its F 3 progeny), between each character pair and
reaction to spot blotch. The nature of the results is illustrated
in Fig. 23, in which the lower Helminthosporium figure indicates
susceptibility and the higher figure, resistance to the disease.
More resistant plants were found in the two-rowed group than
in the six-rowed, in the white than in the black, and in the
rough- than in the smooth-awned. It seemed fair to conclude that
at least three factor pairs, or groups of factors, were involved in
determining reaction to H. sativum, and these were located in
the same chromosomes as the factors for color, row number, and
smooth vs. rough awns. Resistance and susceptibility were not
dependent upon the same factors that conditioned the other
characters since it was possible to obtain a resistant, white-
hulled, six-rowed, smooth-awned variety and also a resistant,
black-hulled variety from the Cross of Svanhals X Lion.
Reaction to St6m Rust Powers and Hines (1933) studied
the reaction to stem rust Puccinia graminis tritici in crosses of
Peatland X Glabron and Peatland X Min. 462. Peatland was
the resistant parent. Glabron and Minn. 462 are sister selections
from a cross of Smooth Awn X Manchuria, and both are sus-
INHERITANCE IN BARLEY
161
ceptible. Reaction to stem rust in the mature-plant stage was
due to a single-factor pair with resistance dominant. Rust
TV
30
20
10
1"
/
7
X
- -* "*
\
<&
\
SN
X N
\
\
f
f
s
\
\
Mean for tffwf'/chon 19.8*0.2
Mean /for vulgoire 17. 2O.J
\
**"^*,
^
I 15 18 21 24 21
Average Helmitrthosporium figure
ou
J
40
f 30
g_20
j"
*E
.* r
/
\
_- - -
-
V
'
\
,/
^Y
$/
' "\
\
\
/
N
^^
\
A/ean /br ^
A/eA7 /br w-
>/ack 17.5*0.2
'hite %0.0 0.2
2 15 18 21 24 2"
12
Helminthosporium f /gure
o> ou
1
30
20
10
I/
\
/
\
*
\
V
\
'
/
\
\
i
Ax'
^^
/
\
Mean ibr'rougft 19.6 0.2
Mean fbr smooth /&0+0.2
*"^
^x.
27
15 18 21 24
Average Helminlhosporium figure
FIG. 23. Distribution of Helminthosporiura reaction of F$ lines homozygous
for other characters in cross of Svanhals X Lion. Lower figure for Hdmintho-
sporium reaction indicates susceptibility and higher figure resistance to the
disease, (from Griffee, 1925.)
reaction was found to be independent of rough vs. smooth awns.
Reid (1938) corroborated the foregoing conclusions regarding
162 METHODS OF PLANT BREEDING
reaction to a large collection of physiologic races of stem rust
in the mature-plant stage in a cross of Barbless X Peatland.
Brookins (1940) found the factor pair for resistance vs. sus-
ceptibility to stem rust (Tt) to be located in the seventh linkage
group, linked with normal vs. brachytic plant type (Brbr) and
normal vs, chlorina seedlings (P\f c ). The gene order was
T 12.6 Br 9.8 F c
16V7
Brookins found also that reaction to physiologic races 19, 36,
and 56 in the seedling stage, in crosses involving Peatland, was
monohybrid, with resistance dominant. The same factor pair
that differentiated seedling reaction to these three races also
controlled reaction to a large collection of races in the mature-
plant stage in the field. Apparently the type of rust resistance
found in Peatland is physiological, since the same gene pair
controlled the same reaction to rust throughout the life of the
plants.
Resistance to Mildew. Stanford and Briggs (1940) sum-
marized the studies on inheritance of resistance to barley mildew
(Erysiphe graminis hordei) carried out at the California Agri-
cultural Experiment Station. From studies of the genetics of
resistance to race 3 in 10 resistant varieties, crossed with one
another and with susceptible Atlas, the factorial composition of
the resistant varieties was as follows:
Variety Factors for Resistance to Mildew
Hanna Mlh Mlh
Goldfoil MlgMlg
Arlington awnless Ml p Ml p Ml y Ml u
Chinerme Ml p Ml p Ml y Ml u
Nigrate Ml p Ml p Ml, Ml v
Algerian Ml a Ml a
S.P.I. 45492 Ml* Mla
Kwan V MlkMk
Psaknon MI P Ml p
Duplex Ml h Ml h MlpMl P ml A mLd
Seven different factors for mildew resistance six dominant
and one recessive were found. The number of resistant factors
in a single variety varied from one to three.
INHERITANCE IN BARLEY 163
The two different factor pairs differentiating Algerian and
Kwan were found to be linked with 9.81 per cent recombination.
The other five factors appeared to be independent of these two
and independent of one another. Thus, seven different factors
are involved in the control of a single physiologic race of a
single disease. This appears to be the largest number yet located
in any species of plants.
INTERACTION OF FACTORS AFFECTING
QUANTITATIVE CHARACTERS
Quantitative characters are extremely important to the plant
breeder. Studies of the genetics of these characters present seri-
ous difficulties, since the number of genes involved usually is
large and the effect of single genes frequently is small. Informa-
tion on the nature of interaction of factors affecting quantitative
characters is very meager.
Powers (1936) studied the nature of the interaction of genes
affecting the four quantitative characters, yield of seed per plant,
number of spikes per plant, height of plant, and length of awn
in a cross between varieties of Hordeum deficiens and H . vulgar e.
Single plants of the parents F\ and F 2 were classified for black vs.
white glumes (Bb), deficiens vs. vulgare type of spike (Ft;), and
normal vs. brachytic type of growth (J3r6r). The yield of seed,
number of spikes, plant height, and awn length were determined
for individual plants. The genotype of the F 2 plants for the
qualitative characters was determined from a progeny test in FS.
Powers found that the homozygous black (BB) and homozy-
gous white (66) segregates did not differ significantly in the
four quantitative characters measured. The heterozygotes (Bb)
exceeded the two homozygotes in all four quantitative char-
acters, although not significantly so in some comparisons. This
increase in the J56 segregates in ^2 over the BB and 66 may be
explained as being due to favorable and at least partially domin-
ant genes located in the chromosome pair carrying J56,
Plants with the vulgare type of spike (vv) yielded more than
those with the deficiens (FF) or the heterozygotes (Ft;). The Vv
segregates yielded more than the FF plants. Normal plants of
the genotypes BrBr or Brbr were higher in yield than the brachy-
tic plants (6r6r)
164 METHODS OF PLANT BREEDING
In making comparisons of the differences in yield of seed
between plants of the genotypes vv Brbr and VV Brbr with vv
brbr and VV brbr, the cross difference (w Brbr VV Brbr)
(w brbr VV brbr) was positive and significant. It is apparent
that the difference in yield between segregates of the vulgare
type (vv) was greater in the presence of the nonallelic genes Brbr
rather than in the presence of the less favorable brbr and of the
deficiens type (VV). In general, it was found that genes favor-
able to high plant yield when transferred from a low to a high
yielding geno-type of a nonallelic factor pair, in comparison with
their alleles, were still more favorable to the development of grain
yield than in the presence of the low yielding genotype.
The foregoing evidence is the reverse of that expected according
to Rasmusson's (1935) interaction hypothesis, which assumes
"that the effect of each factor on the genotype is dependent upon
all the other factors present, the visible effect of a certain factor
being smaller the greater the number of factors acting in the same
direction," Rasmusson found support for his theory in a study
of interaction of factors governing early and late maturity in
Pisum. Powers (1934) in a study of factors governing habit of
growth in Triticum obtained results that support this hypothesis
also.
It is apparent that differences in interaction between genes
controlling quantitative characters occur and that no general
rule can be given at the present time that will describe all condi-
tions. Powers concluded that at the present time any hypothesis
regarding the nature of gene interactions is of doubtful value as
a means of prediction.
CHAPTER XII
INHERITANCE IN FLAX
All cultivated varieties of flax belong to the species Linum
usilatissimum L. The haploid chromosome number usually
found is 15, although 16 in the haploid and 32 in the diploid have
been reported by several investigators (Tammes 1928, Dillman
1936). Tammes gave the chromosome numbers of other Linum
species as 8, 9, 10, 12, 14, 15, and 18 in the haploid condition.
Extensive attempts to cross common varieties of flax with many
of the wild-flax species have been made, but without success
except in crosses between L. usitatissimum and L. angustifolium,
which can be made without difficulty. The hybrids are com-
pletely fertile as a rule. Because L. angustifolium has the same
number of chromosomes and crosses readily with the common
species, it has been considered by Tammes as the probable ances-
tor of common flax. L. angustifolium differs from common
cultivated varieties of flax in that the seeds and capsules are
smaller, the edges of the partition walls of the capsule are hairy,
and the capsules open or dehisce at maturity.
Hairy capsules were dominant to glabrous in FI, and segrega-
tion in F% was on a 3:1 basis. Dehiscence of the capsule at
maturity was imperfectly dominant over the closed type of
capsule, and several factors were necessary to explain segregation
in F%. In a cross of cultivated varieties with a particular form
of L. angustifolium having a strongly tillered and branched habit
of growth of more delicate type of plant than cultivated varieties,
Tammes found no plant among 300 grown in the F% generation
that belonged strictly to the L. usitatissimum type. The length
and width of petal and length of seed in L. angustifolium was less
than that of the cultivated varieties. The inheritance of these
character differences was dependent upon multiple (polymeric)
factors.
There is a similar range of flower colors in L. angustifolium as
in cultivated varieties, although of the factors involved in flower,
165
166
METHODS OF PLANT BREEDING
FIG, 24. For descriptive legend see opposite page,
INHERITANCE IN FLAX
167
seed, and plant colors, only two were exactly the same in L.
angustifolium as in L. usitatissimum.
Factors for Flower and Seed Color in Common Flax. Dillman
(1936) has summarized the major effects of the interaction of
eight genes for seed and flower colors as determined by Tammes.
The pure lines that he obtained from Tammes, their petal,
anther, and seed colors and genetic composition are summarized
in Table 15.
TABLE 15, GENETIC COMPOSITION AND CHAKACTEKS OF PURE LINES
OF FLAX (AFTER DILLMAN)
Description
C.I.
No.
Factor composition*
Petals
Anthers
Seeds
765
AA BiBi BtBz C'C' DD EE FF HH
Blue
Blue
Brown
766
hh
Blue
Yellow
Brown
768
aa
Light blue
Blue
Brown
769
ee hh
Pale blue
Yellow
Brown
770
ff
Lilac
Blue
Brown
771
aa ff
Light lilac
Blue
Brown
772
dd
Pink
Yellow
Brown
773
dd ff
Deep pink
Yellow
Light brown
774
c'c'
White, flat
Blue
Brown
775
bibi
White,
Yellow
Greenish yellow
crimped
776
6*62
White,
Yellow
Brown
crimped
777
bibi c'c'
White, flat
Yellow
Greenish yellow
778
c'c' dd
White, flat
Yellow
Grayish brown
* All dominant factors are present in common blue flax, C. I. 765, the recessive factors
given, being those actually determining the character differences from common blue.
These eight factors are believed by Tammes to be carried in
different chromosomes. From the table it may be noted that
BI, B%, and C f are basic color factors, all in the dominant condition
being necessary for the production of color in the petals. The
FIG. 24. Structure of the flowers of flax.
1. Single flower: (a) calyx; (5) corolla
2. Branch showing: (a) seed boll; (6) calyx; (c) flowers just after blooming; (d) bud.
3. Calyx and corolla removed to show sexual organs in position: (a) anther; (6)
filament; (c) stigma; (d) one of 5 divisions of style; (c) ovary.
4. 6. Cross and longitudinal section of ovary.
5. Ovary, stigma, and 5-lobed style.
7. Cross section of anther.
8. Anther.
Size: 1, about 5 X ; 2, about X ; 3, nearly 4 X ; 4-8, greatly enlarged.
168 METHODS OF PLANT BREEDING
factors D and F determine the tint of the petals. When D is
recessive, in the presence of the basic factors, the petal color is
pink; F in a recessive condition causes lilac; and when both D
and F are recessive, deep pink results. Factors A and E are
intensifies. When a or e are homozygous, recessive, the color
is of a lighter shade.
J5i, -82, C" and D influence the shape of the petals, all four
factors being in a dominant condition in most common flax
varieties with broad, flat petals. If either 61 or 62 are recessive ;
in the presence of both C" and D, the petals are narrow antJ
"crimped," i.e., inrolled at the outer margins. If either C' or D
is recessive, the petals are flat, regardless of the dominant or
recessive condition for BI and J5 2 . The dominant condition of
four factors BI, B 2 , D, and // leads to the production of blue
anthers. When any one of the four factors BI, B^ D, and H is
in a homozygous recessive condition, the color is yellow.
The interaction of two of the genes that influence petal color,
BI and D and a basic factor G for seed color conditions the
development of color in the seed. When G is recessive, the color
of the seed is yellow, because the yellow cotyledons are visible
through the colorless seed coat. There are other factors that
influence the intensity of seed color, and if G is present the seed
may still be yellow. If B i is recessive, in the presence of G and
D, there is a greenish color to the seed. When D or both BI and
D are recessive, the color of the seed is modified from the normal
brown color. Shaw et al. (1931) have postulated the interaction
of at least seven factors that influence the inheritance of petal
color in Indian varieties of flax. Their results are similar to those
of Tammes. Their explanations of the inheritance of seed colors
and crimping of the petals differed materially from those of
Tammes. The genetic factors involved in these Indian varieties
have not been studied in relation to those postulated by Tammes.
Dehiscence of the Bolls. Three types of flax bolls may be
distinguished: dehiscent, semidehiscent, and indehiscent. Most
cultivated varieties of flax in the United States have the semi-
dehiscent type of boll, where the boll opens at the apex and the
five segments separate slightly along the margins. In the inde-
hiscent type is found most of the Indian and Argentine varieties.
The character is of economic importance, since the semidehiscent
types thresh more easily than the indehiscent. Dillman sum-
INHERITANCE IN FLAX 169
marizes crosses made by J. C. Brinsmade, Jr., between the
two types. Semidehiscent was dominant, and ratios in F*
approached 15 semidehiscent:! indehiscent.
Smooth vs. Ciliate Septa. Dillman points out that in most
cultivated varieties of flax the septa are ciliate, although a few
varieties have smooth septa. In most American systematic
botanical statements, the bolls are described as having nonciliate
septa. He credits Brinsmade and A, C. Arny with having
obtained a ratio in F 2 of 3 ciliated:! smooth.
Weight of Seed and Oil Content. Dillman has given the
weight of 1000 seeds in grams for L. angustifoUum and varieties
of common flax grown in 1930 under irrigation at Bozeman,
Montana. The lowest weight was that of the wild species L.
angustifoUum , at 1.5 g. per 1000 seeds, with the heaviest weight
being obtained for the Lino Grande variety of 11.55 g. per 1000
seeds.
Myers (1936) studied seed weight in a cross between Redwing
with a 1000-seed weight of 4.33 g., and Ottawa 770B with a mean
of 5.35 g. There was a partial dominance of large seeds in FI.
One hundred F* lines were grown, and studies of seed weight
were made in comparison with Redwing and Ottawa 770B. One
Fs line had a seed weight nearly as low as Redwing, with a rela-
tively low variance. Two F 3 lines had as great or greater seed
weight than Ottawa 770J3, although the variance for both lines
was significantly greater than that of the parents. In this cross
it is apparent that weight of seed cannot be placed on a simple or
definite factor basis.
Dillman found that the oil content may vary from 33 to 44 per
cent or more, depending upon the interaction of heredity and
environment. Johnson (1932) studied the oil content of 46
varieties of flax grown in replicated rod-row trials at University
Farm, St. Paul, Minnesota, in 1929 and 1930. The material
included both Argentine and domestic varieties and selections
from varietal crosses. The correlations for weight of 1000 seeds
and oil content were +0.72 and +0.78 for the 2 years, respec-
tively. Dillman studied 124 varieties and strains grown at
San Antonio, Texas, in 1926, that ranged in weight of 1000 seeds
from 3.5 to 7.5 g. and in oil content from 36 to 44 per cent. He
obtained a correlation between seed size and oil content of
+0.70,
170
METHODS OF PLANT BREEDING
Dillman placed varieties in four groups on the basis of seed size,
small, midsize, large, and very large. In general, the varieties
with larger seed tend to have higher oil content and selection for
seed size in a cross between parents that differ in seed-size aids
in obtaining varieties with higher oil content.
Inheritance of Quality of Oil. The drying quality of oil is
dependent upon the quantity of oxygen absorbed in the process
of drying to form the characteristic paint film. The chemist
determines the relative drying quality of oils by the absorption
of iodine per unit quantity of oil, the drying quality being
expressed as the iodine number. Values may range from 150 to
200 in extreme samples. Johnson obtained a negative correlat-
tion coefficient of 0.31 for drying quality and weight of 1000
seeds, using 46 varieties grown in rod-row trials in 1930.
Arny (1936) has studied the inheritance of iodine number in
several crosses between common varieties. The character is
influenced rather strongly by environmental factors, individual
plant determinations in the same pure line giving rather wide
ranges in iodine index. Iodine index of parent plants of Bison
and Ottawa 770B and of the Fi and Fa generations are given in
Table 16.
TABLE 16. IODINE NUMBER OF OIL FROM INDIVIDUAL PLANTS OF BISON
AND OTTAWA 770B AND FROM THE Fi AND F z GENERATIONS OF CROSSES
BETWEEN THESE Two VARIETIES
University Farm, 1933*
CO
IQ
05
*Q
S
S
$
r I
t^.
s
i>
j>
8
s
Culture
*
8
4
ci
CD
A
CO
&
JL
>
i
JH
00
Total
Bison
3
3
7
14
6
4
37
770B
fl
14
7
23
Bison X 770BFi
4
3
5
12
Bison X 770BF 2
?,
10
?4
?4
38
37
20
13
6
?,
176
* Unpublished data kindly furnished by A. C. Arny.
In this cross, the FI resembled the low-iodine-index parent,
although dominance was not complete. Segregation occurred
in F%. In backcrosses of the F\ to the parent with higher iodine
index, Arny obtained a 1 : 1 ratio when plants that fell within the
range of the higher iodine-index parent were considered homo-
zygous and those with a lower iodine index were classified as
INHERITANCE IN FLAX
171
heterozygous. These and other data led to the conclusion that a
single factor pair was responsible for the main differences in
quality of oil in most of the crosses studied.
Arny (1936) found rather close linkage between iodine index
and seed color. The following results (cited by Dillman 1936)
are from backcrosses in the coupling phase.
Year
Yellow
v seed
Browi
i seed
Cross
crown
High
Low
High
Low
(Bison: brown, low X C. I. 355;
yellow, medium) X C. I. 355
(C. I. 355 X C. I. 423; yellow,
high) X C. I. 355
1933
1934
64
25
3
7
12
6
56
21
(Bison X C. I. 391; yellow, high) X
C. I. 391
1934
75
4
13
87
Total
164
14
31
164
The calculated recombination percentage of 12.0 1.7 was
obtained.
DISEASE RESISTANCE
Bolley, about 1900, in North Dakota, discovered the organism
that causes wilt in flax and named it Fusarium lini Boll. He was
one of the first to produce an artificial epidemic of a plant disease
as an aid in selecting for resistance. His early work of selection
for wilt resistance in flax and for disease resistance in other crop
plants emphasized the importance and desirability of breeding
for disease resistance.
Wilt Resistance. The severity of infection with wilt is greatly
influenced by environmental conditions, particularly soil tem-
peratures, heritable differences in the degree of resistance of
different varieties, and physiologic races (Broadfoot 1926) of
the pathogen that causes the disease. Tisdale (1916, 1917) made
important contributions to the nature and inheritance of wilt
resistance. High temperature was a favorable agent in overcom-
ing resistance. The fungus penetrates the flax plant through the
stomata of seedlings, the root hairs, or through the young epi-
dermal cells. In a resistant plant, the fungus on entering stimu-
lates cork-wall formation of cells adjacent to those attacked.
172
METHODS OF PLANT BREEDING
Tisdale studied the inheritance of wilt reaction in crosses between
resistant and susceptible strains. Some Fi crosses were much
more resistant than others, and in some crosses there appeared
FIG. 25. Two varieties of flax grown at St. Paul, Minnesota, showing range
in height. Left: Redwing, a typical seed-flax variety. Right: Cirrus, a typical
fiber flax. Varieties of fiber flax are taller and less branched than varieties of
seed flax and produce lower yields of seed,
to be a dominance of resistance, 8 whereas other crosses indicated
a dominance of susceptibility. Although segregation occurred
in JP 2 , the results could be explained only on a multiple-factor
basis.
Bolley (1912) was unable to explain adequately the gradual
accumulation of resistance when the crop was grown on wilt-sick
INHERITANCE IN FLAX 173
soil and the loss of resistance that was often observed after a wilt-
resistant variety had been introduced. He favored the idea that
disease resistance developed by a gradual accumulation of
resistance under infection conditions. He was an early leader
in developing wilt-resistant varieties, and the variety Bison
selected by Bolley is the most widely grown wilt-resistant variety
in the flaxseed area of Minnesota, North Dakota, and adjacent
states. Barker (1923) made a careful study of these problems
and found that some varieties contained no resistant genotypes
and that in these cases resistance was not developed from a
constant association with the pathogen. He found also that
wilt-resistant pure lines did not lose their resistance when grown
on wilt-free soil.
In a study of reaction to wilt, parental lines were self-pol-
linated by Burnham (1932) for at least three generations to ensure
homozygosity. Studies were carried on under field conditions
in wilt-infested soil, and the flax was planted late to ensure
optimum infection. Certain strains of flax used as parents were
completely susceptible; others, highly resistant; but occasional
wilted plants were found in resistant lines used as parents.
These were not believed to be genotypic variations. The extent
of wilting obtained in progeny tests of different inbred parents
ranged from highly resistant through partially susceptible to
highly susceptible. Because of the nature of the disease, plants
for producing F 2 and F 8 progenies were grown on wilt-free soil
and random F 3 lines used to study the genetics of wilt reaction.
From several crosses between resistant and susceptible parents,
the mean for wilting of the F% was generally intermediate between
that of the parents, and F 3 lines were obtained with mean per-
centages of wilting that ranged from high resistance to complete
susceptibility. Approximately 1 out of 10 F 3 lines were as resist-
ant as the resistant parent. In several crosses between resistant
parents, there was a high percentage of susceptibility in F^
indicating that the parents may have differed in factors for resist-
ance. The difficulty of making a genetic analysis is due partially
to the variability in wilt reaction from season to season and to
variations in wilt reaction of the same pure line in different parts
of the wilt nursery. The number and nature of genes responsible
for wilt resistance could not be determined. Even though more
than a single factor pair for wilt reaction was necessary to expla ;i a
174 METHODS OF PLANT BREEDING
the results, it appeared relatively easy, in crosses between resist-
ant and susceptible strains, to recover lines as resistant as the
resistant parent.
Resistance to Rust. The importance of rust resistance in the
seed-growing areas in North America is generally recognized.
In a recent report, Flor (1940) has listed 24 physiologic races
of Melampsora lini (Pers.) Lev. that have been differentiated on
the basis of the reaction of 11 varieties of flax. No varieties of
flax were found that were resistant to all races. Ottawa 770B,
which has been used extensively in studies of the inheritance of
rust reaction, and flaxes of the Argentine type have remained
immune from all races collected in North America. The Argen-
tine selection was susceptible to three physiologic races 19, 20,
and 22 collected in South America, whereas Ottawa 770B
proved susceptible only to race 22. Varieties of flax are available
that are resistant to the South American races of rust but sus-
ceptible to North American races.
Henry (1930) used Ottawa 770B as one of several immune
parents in crosses of immune X susceptible. Immunity was
dominant in Fi. Only a single factor pair was necessary to
explain the crosses between immune vs. susceptible when Ottawa
770B was used as the immune parent. When Argentine selection
was used as the immune parent, segregation was on a 15 : 1 basis.
Myers (1937) grouped parental varieties used in studies of
inheritance of rust reaction into five groups; immune, near
immune, resistant, semiresistant, and susceptible, with the use
of a collection of rust for the source of infection. He explained
results obtained by two allelic series where L and M are duplicate
factors conditioning immunity from the collection, l n and m n
condition near immunity, l r and m r are duplicate factors for
resistance, and I and m are the recessive alleles conditioning sus-
ceptibility. The two series of alleles then would be L, Z n , J r , I,
and At, m n , m r , m. The genotype of Ottawa 770B was considered
to be LL mm.
CHAPTER XIII
METHODS OF SELECTION FOR SPECIAL CHARACTERS
It has been emphasized in preceding chapters that the breeding
of improved varieties almost invariably involves selection for
many characters. Some characters ca& be classified easily, and
visual inspection will determine the desirable ones. Other
characters are difficult to evaluate, and special methods must be
developed to make controlled selection possible. To make the
most rapid progress possible in plant breeding, it is essential that
the breeder cooperate with other plant-science specialists in order
that efficient methods of selection be developed. The methods
used may vary with the nature of the material, the training
of the investigator, and the facilities available. The plant
breeder of today can well afford to make a greater use of plant
physiological technics already available and to take an active
leadership in the development of new technics. Illustrations
will be given of certain technics that have been developed to
aid in character differentiation.
QUALITY TESTS IN WHEAT
A determination of the desirability of a particular variety of
wheat, for human consumption, will depend on the use to be
made of it. The more important uses include bread, macaroni,
pastry, crackers, and breakfast foods. Wheat especially adapted
to one use may be very inferior for some other. The science of
milling and baking is highly specialized, and the satisfactory
evaluation of wheat quality can be made only by the cereal
technologist. This emphasizes again the need of close coopera-
tion between the plant breeder and technologists in other fields.
Among the necessary characteristics of satisfactory bread
wheats is suitable baking strength. This may be defined as the
inherent capabilities of a wheat or flour to produce a loaf of good
volume and satisfactory crumb grain and texture, provided it is
baked under conditions that preclude yeast starvation. Some
175
176 METHODS OF PLANT BREEDING
wheats require special treatment in milling or baking to bring
out optimum results, making it necessary to vary fermentation
time, mixing treatment, and use of " improvers" in order to
obtain a complete evaluation of breadmaking characteristics.
Crumb color in the bread and pigment concentration in the
flour are sometimes important, and so are the dough-handling
properties.
For a discussion of milling and baking methods and tests for
different properties of the flour, the student may be referred to a
book published by the American Association of Cereal Chemists,
"Cereal Laboratory Methods" (1941).
A complete study of the milling and baking properties of a
series of wheats is relatively expensive, and larger quantities of
wheat are required than can be made available during the early
generations in the breeding program. A number of simple,
rapid methods for evaluating certain properties of flour have
been developed. To be of greatest use to the plant breeder, in
selection, such methods must be rapid and inexpensive and must
require relatively small amounts of wheat. None of these
methods alone can replace actual baking trials under commercial
conditions. One such rapid test of baking strength will be
mentioned.
Wheat-meal Fermentation-time Test. The wheat-meal-fer-
mentation-time test originated by Saunders and Humphries
(1928) and developed by Pelshenke (1930, 1933) in Germany and
Cutler and Wor^ella (1931, 1933) in America has aroused a great
deal of interest as a possible means of evaluating baking strength
from small amounts of seed. The method is rapid, inexpensive,
and only 15 g, of wheat is needed. It is of interest to the plant
breeder as a means of selection during the early generations in a
breeding program when large numbers of strains need to be tested
for baking quality. The test is based on the resistance of fer-
menting dough (made from medium finely ground whole-wheat
meal) to disintegration in water. The test consists of making a
ball of dough from the wheat meal, to which a standard amount
of yeast has been added, placing Ihe dough ball in distilled water
in a temperature-controlled incubation chamber, and recording
the time it takes for the dough ball to disintegrate. The time
required for the dough balls to break will furnish a measure of
baking strength, the stronger wheats requiring a longer time.
METHODS OF SELECTION FOR SPECIAL CHARACTERS 177
In general, it may be said that the results of the wheat-meal-
fermentation-time test have been in fair agreement with baking
behavior when wheats differing materially in baking strength are
studied. The test has been used to good advantage, particularly
with soft winter wheats, where soft wheats with a short period of
dough-ball disintegration have been selected. For wheat
varieties that do not differ greatly in baking strength, the meal-
fermentation-time test can not be relied on to differentiate the
varieties according to their reaction in standard baking trials.
This appears to be true particularly with the hard wheats and in
breeding experiments in which both parents have good baking
strength.
COLD-RESISTANCE TESTS WITH WHEAT
Cold resistance of winter-wheat varieties is best measured in
field tests in the region in which the wheat is to be grown. Dif-
ferential killing is obtained only in certain years, and slight
depressions in the field often lead to killing in " patches" with
little relation to varieties. The development of a satisfactory
laboratory test would be of great aid in selecting for cold resistance.
Numerous investigators in the hard-red-winter- and soft-red-
winter-wheat regions of the United States have shown that
artificially produced low temperatures could be used as an index
of the ability of strains of winter wheat to survive in the field.
The methods used and results obtained in a recent study by
Weibel and Quisenberry (1941) will be given briefly.
Thirty varieties of winter wheat grown in the cooperative Great
Plains Uniform Winterhardiness Nursery were used for the study
because of the great amount of information available as to their
relative winter hardiness in the field.
For the test of cold resistance, seed of the varieties was sown
in flats outside during the first week in October (at Lincoln,
Nebraska). Good growing conditions were maintained and the
plants reached the tillering stage before going into a dormant
condition for the winter. Freezing tests were made November
15, December 5, December 15, and January 15 by exposing the
plants to temperature of 17 to 26C. for 24 hr., in a mechani-
cally controlled freezing chamber. After this freezing period,,
the flats were transferred to a greenhouse maintained at 21C
and kept watered to allow live* plants to .recover. Survival
178 METHODS OF PLANT BREEDING
counts were made 10 days after freezing. Twelve replications
were used in each of 2 years.
The interannual correlation between the survival of these 30
varieties in artificial freezing tests in each of 2 years was +.930.
When the average survival, for 2 years, in controlled freezing
tests was correlated with the survival of the same varieties in
the field a coefficient of +.866 was obtained, a highly significant
value.
These wheat varieties differed markedly in cold resistance.
Extensive studies in Minnesota with strains of wheat not differ-
ing greatly in winter survival have shown very little association
between winterkilling under field conditions with reaction in
controlled freezing tests. Winter hardiness involves more than
cold resistance. Alternate freezing and thawing, " heaving " of
the soil, and other factors are of importance also.
SHATTERING IN WHEAT
Most American varieties are relatively resistant to shattering,
probably because of selection for nonshattering during the breed-
ing programs, although marked differences in this character
may be observed. Shattering probably is of greater importance
in the Pacific northwest than in other regions, since wheats in
that region frequently are allowed to stand in the field after
ripening for longer periods of time before being harvested. Wheat
breeders in China have noted that one of the most undesirable
characters of Chinese varieties is their extreme susceptibility to
shattering, as contrasted with introduced varieties (Chang 1940).
Vogel (1938) studied the relation of the amount of mechanical
tissue in the basal portion of the glumes to shattering in wheat.
In general, a direct relationship was found between relative
resistance to shattering and the extent of mechanical tissue at
the breaking point of the glumes. Chang (1940) found a direct
relationship between shattering and the amount of strengthening
tissue in the inner basal portion of the empty glume and the
peripheral region of the basal portion of the lemma.
A simple machine was constructed by Chang to determine
resistance to shattering. This instrument was constructed in
such a manner that turning a crank would cause a rubber paddle
to beat the wheat heads, held on a shattering board, and thresh
a portion of the grain. Three heads were tested in each trial
METHODS OF SELECTION FOR SPECIAL CHARACTERS 179
and 20 trials used for each plot. The mean percentage of shatter-
ing, in different varieties, obtained by the use of this instrument,
varied from 1.3 to 29.6. There was good agreement between
shattering under field conditions and in the controlled studies
with the shattering machine.
DORMANCY IN RELATION TO BREEDING
Afterharvest sprouting may be a problem in the grain fields
of many parts of the world. Varieties are known to vary greatly
in length of the dormancy period after grain is ripe. In certain
regions, the plant breeder wishes to select strains or varieties
that are not susceptible to early germination in order to escape
the losses in yield and in quality from germination of the grain
in the shock as a result of rains after harvest.
Larson et al, (1936) studied the length of the rest period of
common varieties of wheat, oats, barley, and rye by germination
tests at three stages of ripeness: soft dough, hard dough, and ripe.
The rest period was longest in immature seeds. The length of
the rest period varied greatly with the variety. In general,
winter wheats had a shorter rest period than spring wheats.
The spring-wheat varieties Mindum, Marquillo, Kubanka, and
Thatcher were found to have a long rest period.
Harrington and Knowles (1940a) presented the results of tests
of the length of the dormancy period of varieties of wheat and
barley. Head samples were collected from the varieties to be
studied at the stage of maturity when the lower kernels on the
spikes could be indented with difficulty with the thumbnail.
Germination tests were made at intervals of 4, 8, 12, 16, 21, 26,
and 36 days after maturity.
The varieties of spring wheat varied from the inability of
Reliance to remain dormant more than 2 days after maturity
to the ability of Renown to hold a high degree of dormancy for
2 weeks.
In another paper, Harrington and Knowles (19406) concluded
that the failure of the seeds of some varieties to germinate after
exposure to moist weather for several days after harvest is due
to the dormancy period of the varieties and not due to slow
germination. Apex, Thatcher, and Renown were high in resist-
ance to sprouting, whereas Garnet was very low. Chang (1940)
also found afterharvest sprouting to be related definitely to
180 METHODS OF PLANT BREEDING
length of dormancy in hard red spring wheat, barley, and oat
varieties.
Harrington and Knowles found that the variation in amount of
sprouting of strains from crosses was related directly to the
sprouting characteristics of the parents. Transgressive segrega-
tion occurred in some crosses, strains more resistant to sprouting
than either parent being obtained. Tests of the amount of
sprouting immediately after harvest can be used in selecting
strains that are not deficient in this character.
LODGING IN SMALL GRAINS AND CORN
Lodging frequently results in serious losses in yield and quality.
It is a difficult character to evaluate, since it is affected by
numerous characters of the plant and conditions of the environ-
ment. In some seasons little or no lodging is obtained. In other
seasons storms may cause most or all varieties to lodge. This
situation naturally has led investigators to study differences in
plant characters that might be associated with lodging.
Holbert (1924), Hall (1934), and others have found lodging
in corn to be significantly correlated with the force necessary to
pull the plant from .the soil. This has led to determinations of
the pulling resistance of inbred strains and hybrids of corn as a
measure of their ability to withstand lodging.
Salmon (1931) devised an instrument for measuring the
strength of straw of small grains. Strength of straw was meas-
ured in terms of the force necessary to break a given number of
straws. Salmon showed that breaking strength of the straw
was correlated with lodging behavior in the field. These results
have been substantiated by several other investigators.
Atkins (1938) made an extensive study of strength of straw
and other characters in relation to lodging in winter wheat.
Straw strength was based on the force required to break five
straws taken at the first upright internode above the crown of the
plant. Twenty determinations were made per variety.
Relative breaking strength pf straw was fairly constant from
year to year, whereas lodging was not. Average lodging for
several years was correlated significantly with average straw
strength. In a single season, the correlation was not significant.
In view of the variability in lodging from year to year, Atkins
concluded that breaking strength for a single season was a more
METHODS OF SELECTION FOR SPECIAL CHARACTERS 181
reliable index of lodging than was a record of lodging for a single
season.
Clark and Wilson (1933) correlated the average breaking
strength of 30 culms per variety in spring wheat, for a single
test, with the average lodging index determined from rod-row
trials at four stations in Minnesota for 3 years. The correlation
coefficient was nonsignificant. Breaking strength and diameter
of the culms were correlated to the extent of + .537 .148,
DROUGHT STUDIES WITH CORN
Corn yields in the Great Plains region of the United States are
frequently reduced greatly by periods of very high temperature,
low humidity, and deficient soil moisture in midsummer. Under
these conditions, it has been noted that inbred lines and hybrids
vary in their ability to withstand such periods of high tempera-
ture. Seasonal conditions vary greatly, and high temperatures
do not occur every year, making selection for drought resistance
difficult. Consequently a laboratory test would be highly
desirable.
Hunter et al. (1936) and Heyne and Laude (1940), in Kansas,
described a method for testing corn seedlings for resistance to
high temperatures. The corn was planted in 4-in. unglazed
pots, with enough seed to ensure a uniform stand of seven
plants per pot. When the seedlings were from eighteen to
twenty days old, they were placed in a heated room at a tem-
perature of 130F. and a relative humidity of from 25 to 27 per
cent for 5 hr. The plants were supplied with enough water prior
to the test to keep the soil moist throughout the 5-hr. test. The
amount of injury, expressed as a percentage of exposed leaf and
sheath tissue that had been killed, was estimated 3 days after
treatment, and the number of plants killed and degree of recovery
were determined 10 days after treatment.
The reaction of 90 per cent of the inbred lines of corn subjected
to controlled high temperature in the seedling stage was in
accord with the known behavior in the field under extreme
temperatures in midsummer. All lines that were low in resistance
to heat in the field reacted in a similar manner in the seedling
test. The high-temperature test in the seedling stage was con-
sidered a valuable supplement to. studies of drought resistance
in the field.
182 METHODS OF PLANT BREEDING
INDUCING BIENNIAL SWEET CLOVER TO FLOWER THE
FIRST YEAR
Planting sweet clover in the field results in flowering plants
being obtained only in the second year. This makes for slow
progress in breeding.
It has been found that extending the length of day to from 18
to 24 hr. by means of supplementary light stimulates flowering of
seedlings of biennial sweet clover from 6 to 8 weeks after emer-
gence of the seedlings. This makes it possible to grow a genera-
tion in the greenhouse during the winter months. The plants
would be small, however, and selection for type would be
relatively ineffective.
A seed crop the first year on relatively normal plants may be
obtained by planting the seed in pots or flats in the greenhouse
in February and subjecting the seedlings to enforced dormancy
when they are from 2 to 3 in. tall by keeping them at 0C. for
about 20 days. After the cold treatment, the seedlings are
allowed to grow under normal temperatures in the greenhouse
until ready to transplant in the field. Seedlings so treated will
produce fairly large plants the same season and flower profusely.
DETERMINATION OF COUMARIN CONTENT IN SWEET CLOVER
Strains and species of sweet clover, Melilotus, vary greatly in
amount of coumarin, the compound that gives sweet clover its
bitter taste and that has recently been shown by Campbell
et al. (1940) to form the hemorrhagic substance in spoiled sweet-
clover hay. Development of strains of sweet clover low in cou-
marin would improve the palatability of the crop greatly and
reduce danger in its use as hay.
Several methods for the determination of coumarin content in
sweet clover have been developed. The method developed by
Clayton and Larmour (1935) and Stevenson and Clayton (1936)
has been modified slightly in Minnesota and is described for
leaf-tissue analysis here in detail.
Approximately 40 leaves are collected from each plant, in
duplicate, well distributed among several branches. The leaves
are taken from the region 4 to 12 in. from the tips of the main
stem or side branches and with as short petioles as possible. It
is known that the amount of coumarin varies considerably in
METHODS OF SELECTION FOR SPECIAL CHARACTERS 183
different parts of the plants. Consequently, care should be used
in collecting the sample of leaves from the same relative portions
of the plants.
The leaves collected from each plant are mixed, and a 1-g.
sample for extraction and a 1-g. sample for dry-matter deter-
mination are weighed. The dry-matter sample is dried in an
oven kept at 105C. for 20 hr. and the dry weight determined.
The 1-g. sample for coumarin determination is placed in a small
glass vial closed with a cork stopper, frozen as soon as possible,
and kept frozen until ready for analysis. The advantage of
storing the frozen samples is that many plants may be rapidly
sampled, and any plants that may be discarded later on the
basis of disease or plant characters have not been needlessly
analyzed.
In making the analyses, the sample is ground in a mortar with
1 cc. of fine, washed sand. The ground sample is transferred to
a 70-cc. test tube. The mortar and pestle are cleaned during
transfer with 50 per cent methyl alcohol and the sample brought
up to 51-cc. volume with 50 per cent alcohol. The sample is
shaken thoroughly (in a shaking machine) for 30 min. to extract
the coumarin from the ground-leaf tissue. A portion (about
25 cc.) of the extract is filtered, and 5 cc. is placed in a tightly
stoppered vial. This is allowed to stand in light for 12 hr. to
break down the chlorophyll.
The 5-cc. sample of the filtrate is transferred to a test tube
graduated for 50 cc., and 5 cc. of 1.1 per cent Na2COs and 20 cc.
of distilled water are added. The mixture is now placed in a
water bath at 80C. for 15 min., removed, cooled to room tem-
perature; 5 cc. of ice-cold diazonium solution is added, and the
volume made to 50 cc. with distilled water and mixed. The
sample is then allowed to stand for 2 hr., after which the amount
of coumarin is determined through comparison with standard
solutions with known amounts of coumarin.
Twelve standards containing 0.0 to 1.2 mg. of pure coumarin
per 50 cc. are made up. To do so, the requisite amount of
standard coumarin solution is placed in a test tube graduated for
50 cc., about 20 cc. of distilled water, 5 cc. of alfalfa extract, 1 and
5 cc. of 1.1 per cent Na2C(>3 are added. The test tubes are placed
1 The alfalfa extract is prepared by adding 500 cc, of 50 per cent methyl
alcohol to 10 g. of finely sliced alfalfa, shaking for 2 hr., and filtering.
184
METHODS OF PLANT BREEDING
in a water bath at 80C. for 15 min., cooled, 5 cc. of the cold
diazonium solution is added, the resultant solution made up to
50-cc. volume, mixed, and allowed to stand for 2 hr. Readings
on these standards are obtained in a photoelectric colorimeter and
a curve drawn.
The diazonium solution is prepared from two solutions, A and
B, as follows:
Solution A. Dissolve 3.5 g. of p-nitraniline ill 45 cc. of 37 per
cent hydrochloric acid, dilute to 500 cc. with distilled water, and
filter. This solution keeps indefinitely if stoppered.
Solution B. Dissolve 5 g. of sodium nitrite in 100 cc. of dis-
tilled water. Keep this solution in a dark bottle away from light.
This solution should be renewed frequently, since it does not
keep well.
Diazonium Solution. Thoroughly chill a 100-cc. flask, solution
A and solution B, on chipped ice. Pipette 3 cc. of solution A and
3 cc. of solution B into the 100-cc. flask, chill for 5 min., add
12 cc. of solution J5, shake, chill for another 5 min., fill to the
100-cc. mark with ice-cold distilled water, mix, and place on
chipped ice for 15 min. befol-e using. If kept on ice, this solution
will remain stable for at least 24 hr.
TABLE 17. FREQUENCIES OP COUMAKIN PERCENTAGES IN PARENT CHECKS
AND IN F 2 OF CROSSES BETWEEN HlGH AND LOW CoUMARIN SELECTIONS
AND THEIR PARENTS
Parent or F%
Number of plants with coumarin percentage of
0.00
0.01
0.02
0.03
0.04
0.05
0.10
0.20
0.30
0.40
0.50
0.60
Low parent. . .
High parent . .
F z
11
17
25
7
2
12
1
8
1
7
2
4
1
1
19
4
21
10
43
12
35
1
35
The field samples are tested in the colorimeter, the reading
being compared with the curve for the standards. Since a 5-cc.
aliquot of the unknown sample represents 0.1 g. of the original, it
follows that the colorimeter reading for a given standard solution
in milligrams represents the percentage of coumarin in the sample
for analysis. The amount of coumarin is calculated to a mois-
ture-free basis, For coumarin analysis of seed, an incubation
METHODS OF SELECTION FOR SPECIAL CHARACTERS 185
period'is necessary. An alternative micromethod of analysis has
been described by Roberts and Link (1937).
Stevenson and White (1940) reported that through continuous
selection in inbred lines a strain of sweet clover has been devel-
oped with only about one-tenth the amount of coumarin found in
ordinary sweet clover. Stevenson and White crossed low- with
high-coumarin selections and studied segregation in JFV The
results are given in Table 17.
The F% distribution appears to be definitely bimodal. There
were 55 F 2 plants with 0.00 to 0.05 per cent coumarin and 153
with 0.10 to 0.60 per cent. The results indicate that low
coumarin is inherited as a recessive in an apparently simple
manner.
METHOD FOR DETERMINING HYDROCYANIC ACID CONTENT OF
SINGLE PLANTS OF SUDAN GRASS
Individual plants of sudan grass vary greatly in hydrocyanic
acid content. Since HCN is extremely toxic, it is highly desirable
to breed strains of sudan grass free from or very low in HCN
content. If single plants are to be the unit of selection, a test
for HCN must be rapid and relatively inexpensive.
Hogg and Ahlgren, at Wisconsin (unpublished), have used a
procedure based on a method developed by Nowogad and Mac-
Vicar (1940). A description of this method, by Henry L.
Ahlgreif (unpublished communication) is given here.
The method consists of placing .15 grams of green plant material, cut
into short pieces with a scissors or macerated, in a test tube, adding
three or four drops of chloroform, and suspending a strip of moist filter
paper saturated with sodium picrate solution above the mixture. The
saturated filter paper is held in place with a cork stopper which is used
to seal the test tube. The mixture is incubated at room temperature
(20C.) for 12 to 24 hours. The sodium picrate present on the filter
paper is reduced in the presence of hydrocyanic acid. The color is
dissolved out of the paper by placing the paper in a clean test tube
containing 10 cc. of distilled water and is matched with color standards.
The test is sufficiently accurate quantitatively for the selection of plants
low in hydrocyanic acid. The results may be expressed in relative
terms such as "high," "medium" or "low" or in approximate P.PJV1.
based on the percentage of dry matter in the sample.
Tillers from 5 to 7 inches in height can be used regardless of the height
of growth of the remaining portions of the plant. The samples for
186 METHODS OF PLANT BREEDING
analysis for hydrocyanic acid are taken from that portion of the tiller
immediately below the uppermost leaf collar.
The^ reagents and standards are prepared as fdllbws:
The alkaline picrate solution is prepared by dissolving 25 g. of NazCOs
and 5 g. of picric acid in 1000 cc. of distilled water.
Chloroform of Merck's U.S.P, grade is used.
The color standards are prepared by dissolving 0.241 g. KCN in
1000 cc. of water. This gives a stock solution containing 0.1 mg HCN
per cc. Place 5 cc. of the alkaline picrate solution and fSoc, of the KCN
solution in a test tube. Add the following amounts of the KCN-alkaline
picrate solution to eight test tubes
Tube Number Cc. Solution
1 0.00
2 0,10
3 0.20
4 0.40
5 -0.60
6 0.80
. 7 1,00
8 1.60
Bring the volume of each test tube up to 10 cc. by adding distilled
water and heat to boiling in a beaker of water. Permit test tubes to
stand in boiling water for five minutes. Stopper tubes and keep in a
cool place. The number of milligrams of HCN present in each test
tube is as follows: tube 1, 0.00; tube 2, 0.005; tube 3, 0.01; tube 4, 0.02;
tube 5, 0.03; tube 6, 0.04; tube 7, 0.05; and tube 8, 0.08. These stand-
ards can be used for two weeks.
The test paper is prepared by cutting sheets of filter paper into strips
10 to 12 cm. long and 0.5 cm. wide and saturating them with alkaline
picrate solution.
CHAPTER XIV
DEVELOPMENT OF METHODS OF CORN BREEDING
SELECTION WITHOUT CONTROLLED POLLINATION
* * '",
Methods of corn breeding have been studied Extensively since
the introduction of the ear-to-roi? method of breeding by Hopkins
(1899) in 1896. This consisted of growing and studying the
progeny of each ear selected in a single rx>w and continuing selec-
tion from the better yielding rows. Later the method was
improved by replication of rows from each ear and detasseling a
part of the rows from which seed was selected to prevent too close
inbreeding. Williams (1905, 1907) first suggested a remnant
method by which a part of each ear was saved to use for increase
after the better yielding rows had been determined. By this
plan three plots were needed each year, the ear-to-row trial plot,
an increase plot, where the better lines as determined from the
ear-to-row trial were increased, and a multiplication plot from
seed produced the previous year in the increase plot. He also
suggested cooperation and exchange of material among several
breeders as a means of avoiding too close inbreeding. Mont-
gomery (1909) suggested that the ear-to-row plot be used only
once in several years. In intervening years, a seed plot was
planted, and selection of seed was made from vigorous plants in
perfect stand hills.
The usual result from this type of selection is well illustrated by
studies made by Kiesselbach (1922), in which the yields given
represent averages from 1911 to 1917,
Data obtained gave an opportunity to compare the yields of
foiir different methods of seed selection and were as follows:
Average Yield,
Type of Selection Bu.
1. Original Hogue's (without selection.) 53,6
2. Continuous ear-to-row since 1903 53 . 3
3. Increase from single high-yielding strain selected in
1906 , 47,7
4. Increase from composite of four high-yielding strains
selected in 1906 5,0
187
188 METHODS Of PLANT BREEDING
In method 2, the better ears from the highest yielding strains
were selected, whereas in method 3, the remnants of the high-
yielding strain were planted in an isolated plot, and in subsequent
years the better developed ears were selected. Method 4 was an
increase from ear-to-row trials made in 1906 and 1907, where the
four high-yielding ears gave an average yield of 79.4 bu., com-
pared with 64.4 bu. for the original. Subsequent selection was
made in the same manner as in method 3.
It is generally agreed that the ear-to-row method is valuable as
a means of selection with an unadapted variety but, in general, of
little use with an adapted variety. Several studies may be
summarized to emphasize these conclusions.
Hayes and Alexander (1924) compared various methods of selec-
tion in isolated plots, using Rustler White Dent, which had been
grown previously in central Minnesota for many years without
close selection to type. The methods of selection were as follows :
1. Selection of good ears at husking.
2. Selection during seed-corn week in the first half of Septem-
ber, from perfect stand hills and vigorous plants, without close
selection to ear type.
3. Selection as in method 2 and then reselection for ear type,
i.e., good buts, medium dent, straight rows, cylindrical ears,
14 to 16 rows, good ear length.
4. Montgomery's method from 100 ear-to-row plots, where
remnants from the 25 higher yielding ears were bulked and sub-
sequent selection was carried on by method 3.
5. Williams' method, Fi cross of remnants of three better
yielding ears.
6. Multiplication of seed produced by method 5.
The following data are an average of 4 years' results, except as
noted.
Method of Selection Yield, Bu.
1 54.5 0.8
2 54. 3 0.8
3 53. 2 0.7
4 ' 55.2 0.8
5 55.5 0.8
6 96 per cent of method 1.3 years
Comparing methods 2 and 3, by pairing the differences for each
of ttye 4 years, the chances were 37 : 1 that the difference in yield-
DEVELOPMENT OF METHODS OF CORN BREEDING 1?9
ing ability was significant. These data indicate a harmful result
from close selection to ear type.
Smith and Brunson (1925) compared ear-to-row breeding with
simple mass selection in an isolated field of several acres, making
comparative trials over a 10-year period. They started originally
with 990 ears in an ear-to-row trial, selecting the 40 high-yielding
and 40 low-yielding, respectively. The remnants of ears used to
plant the high-yielding rows were mixed to make a high-yielding
group, and the remnants of the low-yielding ears were mixed to
make a low-yielding group. They continued breeding plots
separately for high yield and low yield, selecting 40 ears for each
type of selection, respectively, in subsequent years, and detasseled
alternate halves of rows, saving the seed from the detasseled half
for planting next year. Four ears were selected from each of the
10 highest yielding rows in the high-yielding selections and,
similarly, 4 ears from each of the 10 lowest yielding in the low-
yielding selections. Each of the three methods of selection
simple mass, high, and low yield were carried in isolated plots.
Yield trials were made in another plot with the use of composite
samples of seed. The chances were very great that the high
selection yielded more than the low, but the odds were only
approximately 3 : 1 that the high selection yielded more than the
nonpedigree. Smith and Brunson concluded that continuous
ear-to-row breeding was of little value.
It seems unnecessary here to summarize the extensive experi-
ments that have been made to determine the relationship between
ear characters and yielding ability. In general, the well-known
experiments of Williams and WeltOn (1915) and many others
show no close relationship between ear characters and yielding
ability. The probable reason for these results can be appreciated
by referring to a study reported by Garrison and Richey (1925)
on 'the effects of continuous selection with Boone County White
(C. I. 119). They selected continuously for 6 different ear types
for an 8-year period and compared the yielding ability with
unselected seed of Boone County White. Each type of selection
was made in an isolated seed plot, and mixed seed from at least
50 ears was used to plant each plot. The following types of
selection were made:
1. Strain 1. Rough ears, 8 in. or more in length, with 20 or
more rows of crease- to pinch~dent#d kernels,
190 METHODS OF PLANT BREEDING
2. Strain 2. Rough ears, 8 in. or more in length, with 16 rows
of crease- to pinch-dented kernels.
3. Strain 3. Smooth ears, 10 in. or more in length, with 20 or
more rows of dimple to slightly crease-dented kernels.
4. Strain 4. Smooth ears, 10 in. or more in length, with 14
rows of dimple-dented kernels.
5. Strain 5. Smooth ears, 10 in. or more in length, with 12
rows of dimple-dented kernels.
6. Strain 6. Smooth ears, any length, with eight rows of
dimple-dented kernels. This strain originated from a few eight-
rowed ears found among those in strains 4 and 5 in 1918.
Selection was effective, since row numbers were rather rapidly
modified. The following quotation taken from Garrison and
Richey (1925) gives a good idea of the more important results and
conclusions.
Without regard to the reason, it is evident that close selection to any
type, as practiced in these experiments, resulted in decreased productive-
ness. The most productive strain, No. 4, the 14-rowed smooth selec-
tion, yielded 8.4 0.20 per cent less than C. I. No. 119, and the least
productive, No. 3, the 20-rowed smooth selection, 14.3 0.19 per cent
less. The 14-rowed smooth and the 16-rowed rough selections, Nos. 4
and 2, were more productive and also departed less from the character-
istic condition of the parent variety than the others.
In their practical application the experiments indicate that a decrease
in vigor and productiveness similar to that followiiig inbreeding may
result from too close selection for a particular kind of ear. Careful
experiments have failed to demonstrate a marked consistent superiority
for any specific kind of ear. Other experiments have shown that the
yields of crosses between varieties of corn frequently are more productive
than the average of the parents, thus indicating that the parent varieties
are too homozygous to permit maximum yields. Just what constitutes
too close selection is not known. In view of the lack of evidence in
favor of any particular kind of ear and the abundant evidence of the
decreased yields that follow close breeding, however, it seems best to
stay on the safe side by avoiding such close selection.
In view of the lack of evidence of marked consistent superiority for
any particular kind of ear, it is finfortunate to teach that uniformity
among the ears of a variety of corn is desirable by attaching importance
to uniformity of sample, as is done in corn shows.
These and many other similar studies show that mass- or
individual-plant methods of selection with an adapted variety
DEVELOPMENT OF METHODS OF CORN BREEDING 101
cannot be expected greatly to increase the potential yielding
ability of the variety, and for this reasdfi plant breeders have
rather generally adopted controlled pollination methods of
breeding during the last 15 years. The student who may wish to
make a more extensive study of earlier methods of corn breeding
is referred to a paper written by Richey (1922) that reviews the
studies in considerable detail and contains, also, citations of the
more important literature,
The conclusions to be reached from these studies is that simple
mass selection for vigor of plant, time ctf, maturity, etc*, is all that
is worth while and that close selection for ear type is not desirable.
These conclusions have aided in a rather rapid acceptance of
modern methods.
EARLY STUDIES OF SELF- AND CROSS-FERTILIZATION
WITH CORN
Although Beal, in 1876, at Michigan, suggested the use of Fi
crosses between strains of corn for a commercial crop and
Morrow and Gardner, in 1892, in Illinois, gave an outline of a
method for producing FI seed and presented further data regard-
ing its value, the utilization of hybrid vigor in corn has been
developed only after extensive experiments on the effects of
cross- and self-fertilization. Studies by East, at the Connecticut
Experiment Station, and of Shull, at Cold Spring Harbor, have
been discussed under the heading of Heterosis. Both started
studies of the effects of self-pollination in corn in 1905. One of
the present writers worked under East's direction in 1909 and had
charge of the corn-breeding program at the Connecticut station
from 1910 to 1914. Part of the self-pollinated lines started by
East, in 1905, have been continued at the Connecticut Agricul-
tural Experiment Station until the present time, being carried on
by D. F. Jones since 1915. Because of the relation of principles
learned with corn to the breeding of other cross-pollinated crops,
it may be desirable to mention other early investigators who have
studied self- and cross-fertilization with corn. G. N. Collins, of
the U.S. Department of Agriculture, who was interested chiefly
in fundamental principles, published his first paper on corn
breeding in 1909. Vice-president Henry A. Wallace started
studies of self-pollination and selection in 1913 as a private enter-
prise, This led eventually to the formation of the Pioneer
192 METHODS OF PLANT BREEDING
Hi-Bred Corn Company of Iowa. Self-pollination and selection
with corn was begun *at the Minnesota station in 1914. F. D.
Richey, who began self*pollination of corn in 1916, was placed in
charge of corn improvement for the U.S. Department of Agricul-
ture in 1922, and his leadership was responsible to a considerable
extent for the rapid development of hybrids adapted to the corn
. belt. Studies of selection in self-pollinated lines were started also
in 1916 by C. H. Kyle and J. R. Holbert as a part of the program
of the Bureau of Plant Industry. Corn improvement was placed
on a cooDerative basis by corn-belt experiment stations and the
U.S. Department of Agriculture under the Purnell Act in 1925.
A committee was appointed to formulate a program. Annual
meetings in the field and laboratory furnished a medium for the
exchange of ideas and materials and without doubt were responsi-
ble to a considerable extent for the rapid development in recent
years of adapted hybrids for all sections of the corn belt.
A summary of conclusions drawn by Shull (1910) from his
studies of self- and cross-fertilization show the detailed knowledge
that was available many years ago. The following summary is
quoted from Shull.
1. The progeny of every self-fertilized corn plant is of inferior size,
vigor, and productiveness, as compared with the progeny of a normally
cross-bred plant derived from the same source. This is true when the
chosen- parent is above the average conditions as well as when below it.
2. The decrease in size and vigor which accompanies self-fertilization
is greatest in the first generation, and becomes less and less in each
succeeding generation until a condition is reached in which there is
(presumably) no more loss of vigor.
3. Self-fertilized families from a common origin differ from one
another in definite hereditary morphological characters.
4. Regression of fluctuating characters has been observed to take
place away from the common mean or average of the several families
instead of toward it.
5. A cross between sibs within a self -fertilized family shows little
or no improvement over self-fertilization in the same family.
6. A cross between plants belonging to two self-fertilized families
results in a progeny of as great vigor, size, and productiveness, as are
possessed by families which had never been self-fertilized.
7. The reciprocal crosses between two distinct self-fertilized families
are equal, and possess the characters of the original corn with which
the experiments were started.
DEVELOPMENT OF METHODS OF CORN BREEDING 193
8. The FI from a combination of plants belonging to certain self-
fertilized families produces a yield superior to that of the original
cross-bred stock*
9. The yield and the quality of the crop produced are functions of
the particular combination of self-fertilized parental types, and these
qualities remain the same whenever the cross is repeated.
10. The FI hybrids are no more variable than the pure strains which
enter into them.
11. The Fa shows much greater variation than the FI.
12. The yield per acre of the Fife less than that of the FI.
Effects of self-fertilization were discussed in somewhat greater
detail by East and Hayes (1912). The following statements
summarize the more important results of inbreeding and selection.
1. Loss of vegetative vigor has followed continued self-pollina-
tion in all inbred lines of corn.
2. Inbred lines exhibit differences in many normal characters;
for example, some inbred lines have long ears; others, short ears.
3. Some inbred lines are much more vigorous than others, even
though they do not differ in degree of homozygosity.
4. Some pure strains are so lacking in vegetative vigor that
they cannot be propagated.
5. Continued inbreeding leads to purity of type.
Shull (1909) outlined a pure-line method of corn breeding based
on the isolation of self-pollinated lines and the use of FI crosses
between them for the commercial crop. The difficulty of this
method was the cost of seed from single crosses.
The work of Jones, at the Connecticut Experiment Station,
since 1915 was continued with some of the inbred lines available
from the early work of East and Hayes. The Mendelian eiplana-
tion of hybrid vigor, given by Jones (1917) was of great value in
a.n understanding of corn-breeding principles. The double-cross
plan of corn breeding, developed by Jones about 1917, has helped
materially to make hybrid seed production economically feasible.
CONTROLLED POLLINATION METHODS
Many workers have taken part in the extensive studies that
have Jed to a partial standardization of modern methods of corn
breeding. Some of the investigations will be briefly summarized.
Although individual research has been the basis of a standardiza-
tion of technic, the rather rapid acceptance of methods has been
194
METHODS OF PLANT BREEDING
brought about by cooperation and free exchange of ideas and
material among investigators.
The following brief summaries of investigations are given to
furnish a background for an understanding of the development of
methods of corn breeding.
Two major problems are faced by the corn breeder. There are
(1) the isolation of the most desirable inbred lines and (2) the use
of these lines to produce hybrids with high yielding ability that
excel also in other characters.
One of the problems not entirely solved yet and one that per-
haps never will be solved is the extent of homozygosity necessary
in inbred lines. Richey and Mayer (1925) made a comparison
of crosses between inbred lines after 3 and 5 years of self-fertiliza-
tion and concluded that there was no general advantage in yield-
ing ability from crosses between lines inbred for 5 years over
crosses between lines inbred for 3 years. They found little or no
relationship between the productiveness of self ed lines and that of
their crosses. Richey (1924) and Richey and Mayer (1925)
found that certain lines behaved rather uniformly in different
crosses and that, on the average, certain strains gave high-yielding
crosses when combined with a random series of other inbred lines ;
i.e., certain lines were good combiners.
At Minnesota, Nilsson-Leissner (1927) and Jorgenson and
Brewbaker (1927) compared the yielding ability of inbred lines
and of possible F\ crosses between them by means of the correla-
tion coefficient. The number of lines, source of material, place of
study, and the correlation coefficients are given in the following
summary :
Source of
material
Number of
inbred lines
Place
Correlation
coefficient
Flints
9
University Farm, St. Paul,
+ .74 .04
Dents
13
Minn,
University Farm, St. Paul,
+ .19 06
Silver King
10
Minn.
Waseca, Minn.
+ .50 .08
Multiple correlation coefficients were calculated also between
five characters of the inbred lines and the yield of their Fi crosses
DEVELOPMENT OF METHODS OF CORN BREEDING 195
as follows: University Farm, Dents, R = .67; Flints, R = ,82;
Waseca, Silver King, R = .61.
In these studies, some lines tended to give good yields in
crosses, and others generally were low combiners. Although
some high-yielding inbred lines did not combine well in general,
the more vigorous inbred lines, on an average, were better
combiners than the less vigorous inbred lines.
Jenkins (1929) made a more detailed and extensive study of a
similar nature. In a study of the relationship of the yielding
ability of inbred lines and their F\ crosses, he found, on an
average, less association than in the Minnesota studies. Coeffi-
cients of correlation in 1926 and 1927 for the mean yield of inbred
lines and their P\ crosses were +.20 .03 in both years. When
the mean of several crosses was used as a criterion of the combin-
ing ability of the inbred lines and correlated with the yield of the
inbred lines, coefficients of +.32 .07 and +.12 .09 were
obtained in 1926 and 1927, respectively. When the mean yield
of the cross-bred progeny was correlated with four characters of
the inbred lines, indicating plant vigor and size, a multiple
correlation coefficient of +.42 .05 was obtained.
Kiesselbach (1930) has shown that advanced-generation
crosses produced, on an average, as high yields in double crosses as
when Fi crosses are used as parents for the double cross. An
advanced generation is the product of normal uncontrolled
pollination in the progeny of an F\ or later generation cross grown
in an isolated plot. One difficulty of using advanced-generation
crosses for producing double-crossed seed, as pointed out by
Kies^elbach, is the reduction in seed production of Ft and F$ in
comparison with F\. His F% and F 3 generations averaged about
67 per cent as much grain yield as the FI. Richey et al. (1934)
found that the F% of 10 double crosses yielded 5 to 24 per cent less
(average 15.2) than the FI. Neal (1935) obtained an average
yield in F 2 and F 3 for advanced-generation single crosses of 70,5
and 75.7 per cent of the F\, respectively. Various investigators
have suggested that advanced-generation seed be used as the male
parent in the commercial crossing plot.
After obtaining as desirable inbred lines as possible, the prob-
lem remains of how to use these inbred lines in hybrid seed
production. Several methods haye been put in practical use.
These include :
METHODS OF PLANT BREEDING
1. Single crosses.
2. Inbred-variety crosses, sometimes called top-crosses.
3. Three-way crosses.
4. Double crosses.
5. Multiple crosses. _
Single crosses can be used only when the inbred lines yield
sufficiently well to make seed production economical. The
higher yielding inbred line should be used as the female parent.
.. .
FIG 26 At left, Cl, an inbred line of Japanese hull-less popcorn; at right, C6,
another inbred iu ceAter, the f i cross, Minhybrid 250. The mbreds yield
sufficient ywcU so that the use of F l crossed seed for commercial planting ;
feasible The F. cross yields approximately 16 per cent more than Japanese
hull-less and has approximately 29 per cent greater popping expansion.
Single crosses are being used in producing sweet corn for canning,
where uniformity of maturity and for ear characters are of major
importance. If the inbred lines are reduced only to practical
homozygosity by 3 or 4 years of inbreeding, single crosses in
sweet corn are economically feasible. An illustration of inbred
lines and the commercial Fj cross of Japanese hull-less popcorn is
given in Fig. 26.
Top-crosses, or inbred-variety crosses, may be expected to
yield somewhat less than single, three-way, or double crosses, on
an average. They have been used 'extensively in sweet-corn
breeding because of their practical features. Before the best
single, three-way, or double crosses have been determined, it
frequently may be of value to use inbred-variety crosses, since
such crosses, when carefully selected, may have a higher yield and
greater uniformity than standard varieties.
DEVELOPMENT OF METHODS OF CORN BREEDING 197
In a three-way cross, a good pollen-producing inbred line is
used as the male parent, and a single cross is used as the female
parent. Both the inbreds used in the single cross should combine
well with the pollen parent. An illustration of the parent inbred
lines, FI cross and three-way cross, Minhybrid 301 is given in
Fig. 27.
FIG. 27. The three-way cross at the right, Minhybrid 301, produced by
crossing the Fi cross, upper center, of two inbred lines of Minn. 13 with an
inbred line of Reid's Yellow Dent, lower center. At the left are representative
ears of inbred lines 14 and 11 of Minn. 13.
In a double cross, two single crosses are used as the parents.
Advanced-generation single crosses can be used to advantage as
the male parents. This is valuable as a means of extending the
pollen-sheddjng period of the male parent.
One of the major problems of the corn breeder is to devise some
method of selecting the better crosses. Richey and coworkers
advocated a series of inbred testers as a means of selecting inbred
lines that were good combiners, determining the combining
ability of each inbred line by crossing it with each of the testers
separately. Jenkins and Brunson (1932) have compared differ-
ent methods of testing the combining ability of inbred lines. In
one case they compared two methods, using 37 inbred lines. The
combining ability was tested in a series of crosses of each of the
37 lines with 9 tester inbred lines in 1927 and in an inbred-variety
cross in 1929. The combining ability of the 3.7 lines determined
by yields obtained by these two methods was correlated, obtain-
ing a correlation coefficient of +-53, where .32 represented a
significant relationship based on odds of 19:1.
198
METHODS OF PLANT BREEDING
In a second study, they used 12 early inbreds in a series of
crosses with 9 comparable inbred lines and 17 late inbreds in a
series of crosses with 10 comparable inbred lines, correlating the
average combining ability of each line with its combining ability
in top- or inbred-variety crosses. The results were as follows:
Lines
used
Correlation coefficient
12 early
17 late
Calculated r, , ...
+ .80
+ .65
Significant r
.58
.48
In a third group, the combining ability in top-crosses and in a
series of 10 comparable single crosses was determined with the
following results :
Lines used
Correlation coefficient
10 early
white
17 early
yellow
10 late
yellow
10 Lancaster
Calculated r
Significant r. . .
+ 90
.60
+ .86
.48
+ .63
.63
+ .90
.63
In a fourth group, GO inbred lines of Pride of Saline were placed
in 6 groups of 10 lines each, and all possible crosses were made
between each line in the odd group with each 10 in the next
higher even group. They determined the combining ability of
each line also by the top-cross method and compared the results
of the two trials, i.e., combining ability in top-crosses with the
combining ability obtained from an average of single crosses, for
each of the 60 lines, obtaining an r of +.56 where a significant r
was .26.
To determine whether crosses with a series of inbred lines used
as testers are more satisfactory as a measure of combining ability,
they studied the combining ability of inbred lines in two different
series of inbred crosses. They used the following groups: (1) 9
early white endosperm inbreds, (2) 9 early yellow, and (3) 10 late
yellow lines. In test A, the combining ability of each line was
DEVELOPMENT OF METHODS OF CORN BREEDING 199
determined from crosses with every other line in the same sub-
group. In test B, the following studies were made of combining
ability. The 9 early white lines were crossed with 13 other inbred
lines of white endosperm; the 9 early lines of yellow were each
crossed with 12 other inbred lines of yellow endosperm, and each
of the 10 late lines were crossed with 17 other late inbreds.
Correlating the combining ability in tests A and B gave the
following :
Correlation coefficient
Groups
9 early
white
9 early
yellow
10 late
Calculated r . ....
Significant r . ...
+ .82
.67
+ .69
.67
+ .65
.63
These coefficients are about the same magnitude as those
obtained by correlating the yielding ability in top-crosses with
that of a series of inbreds used as testers.
The results justify the use of inbred-variety crosses to deter-
mine the combining ability of inbred lines.
Johnson and Hayes (1936) studied the combining ability of
1 1 inbred lines of Golden Bantam in all possible single crosses arid
in top-crosses. The correlation between the average yield of the,
1 1 lines in all possible single crosses and the average yield in top-
crosses with Golden Bantam and Del Maiz was +.78 .12.
The combining ability of 39 lines in two series of top-crosses was
studied also. Johnson and Hayes suggested the need for many
replications and of making yield trials at several locations, prefer-
ably, to determine accurately in a single year the combining
ability of inbred lines by the top-cross method. This plan has
been adopted as a standard practice in the Minnesota corn-
breeding program.
After selecting the more desirable inbred lines by the top-cross
method, it is necessary to determine the best combination of lines
for single, three-way, and double crosses. An actual field trial
must be used to determine the value of a particular combination,
but methods of prediction based on a previous knowledge of
combining ability may be of value.
200
METHODS OF PLANT BREEDING
For a three-way cross between lines, a good pollen producer
must be used as the male parent. If inbred lines 1, 2, and 3 are
used in a three-way cross and inbred line 3 seems most satisfactory
as a pollen parent, it is then necessary that single crosses 1X3
by 2X3 both be good producers in order that the three-way
cross (1 X 2) X 3 be desirable. Thus, a three-way cross may be
-^^^^^^^^^^^^^^^MMaMMBMmaffiaBaBllllilliiiimiliHiliiii \\mmm\mw\m?*
FIG. 28. Representative ears of inbred lines used in double crosses adapted to
south central Minnesota, J^J normal.
selected on the basis of single crosses and a knowledge of the
desirability of the lines as pollen parents.
In a double cross of four lines, i.e., (1 X 2) X (3 X 4), where
single crosses 1X2 and 3 X 4 are planted alternately in. a plot
for hybrid seed production, it would seein desirable to use the
most satisfactory cross 1 X 2 or 3 X 4 as the female parent,
giving consideration to yielding ability and size and uniformity
of the seed of the two FI crosses. Inbred lines used in double
crosses adapted to south central Minnesota, the two FI crosses
used as parents, and the double cro&, Minhybrid 502, are given in
Figs. 28 and 29.
Some studies have been made that furnish a logical basis for the
prediction of the performance of double crosses and that aid in
selecting the more promising combinations for yield trials.
DEVELOPMENT OF METHODS OF CORN BREEDING 201
Jenkins (1934) has presented a study in which 11 inbred lines
were used. From these 11 inbred lines, all but 2 of the possible
single-cross combinations were obtained, and the combining
ability of each of the lines was also tested in an inbred-variety
cross. Forty-two of the possible double crosses of these inbred
lines were studied also, and four methods of predicting the
FIG. 29. Representative ears of single crosses and of the double cross, Minhybrid
502, adapted to south central Minnesota, % normal.
probable value of these 42 double crosses were compared. These
methods are as follows :
1. The yield of all six single crosses from the four lines used in
each double cross was used as a basis for prediction of the prob-
able value of the double cross. If the lines are 1, 2, 3, and 4, these
single crosses are 1 X 2, 1 X 3, 1 X 4, 2 X 3, 2 X 4, and 3X4.
2. ' The yield of four single crosses, excluding the two used as
parents, was used to estimate the probable performance of the
double cross. If the double cross was (1 X 2) (3 X 4), the four
single crosses selected to predict yielding ability would be 1 X 3,
1 X 4, 2-X 3, and 2X4.
3. The mean yielding ability of each line in all possible single
crosses was first determined, and from these means the combining
ability in a particular double cross was estimated. In double
cross (1 X 2) (3 X 4), the probable value of the double cross was
202
METHODS OF PLANT BREEDING
estimated by averaging the combining ability of lines 1 , 2, 3, and 4
in all possible combinations with the other inbred lines. By this
method ; the prediction value for four lines in a double cross
would be the same for (1 X 2) (3 X 4) as for (1 X 3) (2 X 4) or
other combinations of single crosses of these four lines.
4. The combining ability was determined of the four lines in
each double cross by an average of their yields in inbred-variety
crosses.
The value of these four methods of prediction was determined
by correlating yields obtained from each of the prediction
methods with the yield of 42 double crosses, with the following
result :
Predicted yield with
actual yield of
Correlation coefficients by methods
A
B
C
D
Significant
42 double crosses ....
+ 75
+ .76
+ 73
+ .61
39
Although prediction of the value of the double cross by method
D, the inbred-variety cross plan, gave a lower value of r than
methods A , 5, and C, it is possible that this may have been due to
insufficient replication in D or to a chance deviation rather than
to the fact that this method is the least desirable or reliable of any
of the four methods tested.
In discussing method B, Jenkins stated, "In any double cross
the genes of each of the four parental lines are united only with
allelomorphs of the two lines which entered the double cross from
the opposite parent." Extensive studies at Minnesota have
shown that this method can be used to predict the yield of a
double cross about as accurately as by testing the actual double
cross. Doxtator and Johnson (1936) compared the yielding
value of inbred lines in double crosses on the basis of the way that
they are combined and have proved clearly that it is of extreme
importance how these lines are combined, i.e., which combinations
of single crosses are used as parents for the double cross. In
general, other things being equal, the two lower yielding single
crosses of the possible six should be selected as parents of the
double cross.
DEVELOPMENT OF METHODS OF CORN BREEDING 203
The results of these studies are summarized in Table 18.
The predicted yields were obtained by averaging the actual
yields from single crosses. Thus, the predicted yield of (62 X
67) X (66 X 68) was obtained by averaging the yields of the
single crosses (62 X 66), (62 X 68), (67 X 66), and (67 X 68),
which were 58.8, 17.1, 58.9, and 38.8, respectively. This gave a
predicted average yield of 43.4 bu. In general, the agreement
between predicted yields and actual yields obtained was very
close in these studies.
TABLE 18. ACTUAL AND PREDICTED YIELDS OF DOUBLE AND THREE-WAY
CROSSES
Yield, bu. per acre
Hybrid
Obtained
Predicted
Wascca branch station:
(11 X 14) X (374 X 375).. . .
78.70
85.55
(11 X374) X (14 X 375) .. ..
66 33
70.79
(11 X 375) X (14 X 374). .
70.58
69.31
(11 X 14) X374
81 67
86 92
(11 X 14) X375..
82.63
84.17
University Farm:
(62 X 67) X (66 X 68) . .
48 4
43.4
(62 X 66) X (67 X 68) . .
41.8
41.7
(64 X 66) X (62 X 68)
54.1
47.5
(64 X 68) X (62 X 66)
44 5
39.8
Data from Anderson (1938) are presented to give further
information regarding actual and predicted yields. The results
in Table 19 give yields of single crosses in bushels per acre and
show the method of predicting the yield of an actual double cross.
A comparison of actual and predicted yields of double crosses
by Anderson (1938) is given in Table 20 to show the value of the
method.
During recent years extensive unpublished data at Minnesota
show the accuracy of predicting the yielding ability of double
crosses and lead to the conclusion that such predicted yields can
be used as actual measures of the yielding ability of new double
crosses. By comparing the single crosses used in the predictions
with standard double crosses of known yielding ability, one can,
with a high degree of accuracy, accept the predicted yield and use
204
METHODS OF PLANT BREEDING
TABLE 19. METHOD OF PREDICTING YIELDS OF THE THREE DIFFERENT
DOUBLE CROSSES THAT CAN BE MADE FROM FOUR INBRED LINES WITH
THE USE OF THE YlELDS IN BUSHELS PER ACRE OF ALL SlX POSSIBLE
SINGLE CROSSES
(23 X 24) X (26
X 27)
(23 X 26) X (24
X 27)
(23 X 27) X (24
X 26)
Single cross
Yield,
bu.
Single cross
Yield,
bu.
Single cross
Yield,
bu.
(23 X 26)
(23 X 27) ....
(24 X 26) ...
(24 X 27)
62.6
70.8
65.6
72 1
(23 X 24) . . .
(23 X 27) .
(26 X 24) ....
(26 X 27)
41 7
70 8
65 6
64.2
(23 X 24)
(23 X26)
(27 X 24)
(27 X 26)
41.7
62.6
72.1
64 2
Average ....
67 8
Average
60 6
Average
60 2
TABLE 20. A COMPARISON OF ACTUAL YIELDS OF Six DOUBLE CROSSES
WITH PREDICTED YIELDS OBTAINED BY AVERAGING THE YIELDS OF THE
FOUR SINGLE CROSSES NOT USED IN MAKING THE DOUBLE CROSS
iJilieo UUIIILHIIUU. UI1U Ul/UUiU L'iUWM
Actual
Predicted
23, 24, 26, 27:
(23 X 24) X (26 X 27) ....
68 8
67 8
(23 X 26) X (24 X 27)
(23 X 27) X (24 X 26)
23, 24, 26, 28:
(23 X 24) X (26 X 28)
62 4
62.0
65
60.6
60.2
65 5
(23 X 26) X (24 X 28)
59.8
58.0
(23 X 28) X (24 X 26)
56
58 5
23, 24, 27, 28:
(23 X 24) X (27 X 28)
71 1
69 2
(23 X 27) X (24 X 28)
58 1
59 4
(23 X 28) X (24 X 27)
58
60 4
Difference for significance ui 5 per cent
level
5.3
3.4
Bu. per acre
it in the same manner to determine the value of a double cross as
if the actual yield of the double cross had been obtained. The
value of the method is emphasized by giving formulas that show
how many single and double crosses can be produced from n
inbred lines,
DEVELOPMENT OF METHODS OF CORN BREEDING 205
The number of single, double, and other crosses that can be
made from n inbreds may be calculated from the number of
combinations of n inbreds taken r at a time, where r is the number
n\
of inbreds in the cross. This is given by -77 : r-.-
J r\(n r)!
AND THEIR FiRST CROSSES ;
AU6
A 14ft
AW)
FIG. 30. Above, representative ears of four inbred lines; center, the two
single crosses used as parents to produce the double cross; below, representative
ears of four single crosses used to predict double-cross yields. The double cross,
produced by crossing the single crosses given in the center of the picture, matures
satisfactorily in the corn-growing regions of the Minnesota Red River Valley
and has good agronomic characteristics for this region.
For single crosses, r = 2, and the number of possible single
-11 i n (n ~ 1)
crosses will be ^ -
For double crosses, r = 4, and the formula becomes three
nl
times 777 ,.,> since three double- crosses can be made from
4![n 4)!
206 METHODS OF PLANT BREEDING
any four inbreds. This formula may be expressed as
3n(n - \)(n - 2)(n - 3)
24
Thus, for 20 inbred lines, 1 90 different single crosses can be made,
and from the yield trials with these one can predict the actual
yielding ability of 14,535 double crosses.
BREEDING IMPROVED INBRED LINES
A problem of great importance to the breeder is the develop-
ment of improved inbred lines. The following methods are
available at present:
1. Inbreeding and selection from commercial varieties.
2. Inbreeding and selection from high-yielding crosses.
3. Breeding improved lines by a definite plan of crosses and selection.
a. Pedigree methods, as with small grains.
b. Backcrossing.
c. Convergent improvement,
It seems unnecessary to describe these methods in great detail.
Inbreeding and selection from commercial varieties has been the
common practice and is necessary as the first step. As with
other plant-breeding studies, it is important to use as large num-
bers as possible and to determine the breeding value of a line by
progeny trials. A plan in common use is to grow a short row of
30 or more plants from each desirable inbred ear from the previous
generation and to continue inbreeding and selection until practi-
cal homozygosity has been reached. Jones and Singleton (1935)
have suggested growing only a few plants from each inbred ear,
thus making it possible to grow several thousand inbred lines per
acre, with the view that the more important differences will be
those between lines rather than selection within lines. Jenkins
(1935) studied the combining ability in top-crosses with Krug of
14 inbred lines of lodent and 14 of Lancaster for 8 inbred genera-
tions. It was shown that the inbred lines established their
individuality as combiners early and maintained it during succes-
sive generations of selfing. It was suggested that the selection of
desirable combining lines should be determined by crosses made
during early generations of selfing.
As with other methods of plant breeding, it is important to
know the characters desired and carrv on breeding studies with a
DEVELOPMENT OF METHODS OF CORN BREEDING 207
definite plan. If these characters are present in the inbred lines
used in various types of crosses, then these crosses may be a
promising basis as material for developing a new series of inbred
lines. In some cases, this will lead to the pedigree method of
breeding that has been outlined already for self-fertilized crops.
THE PEDIGREE METHOD OF SELECTION IN THE SEGREGATING
GENERATIONS AFTER CROSSING INBREDS
Such a method has been used extensively at Minnesota, and its
value has been discussed by Hayes and Johnson (1939). Most
inbreds obtained from varieties adapted to Minnesota lacked
ability to withstand lodging, and at least one inbred parent of the
single crosses used as a basis for selection of inbred lines was
outstanding in ability to withstand lodging. Selection was made
during the segregating generations from selfed progenies for
ability to withstand lodging, for smut resistance, and for other
desirable characters. Many inbred lines were isolated that
excelled in standing ability and in other characters. The inbred
lines obtained in this way, together with the inbred parents used
in single crosses, were tested for combining ability by crossing
with Minn. 13 and making the necessary yield trials. When
both inbred parents of a single cross were high in combining
ability, as determined from yield trials of inbred-variety crosses,
then practically all inbreds isolated from this particular cross also
showed high combining ability. Conversely, inbreds selected
from a single cross between two lines of low combining ability
were mostly of low combining ability when tested in inbred-
variety crosses, whereas inbreds selected from a cross of a low
combiner inbred with a high combiner gave a range in combining
ability from low to high. These data show that combining
ability is an inherited character. Twelve characters that for the
most part represent vigor of growth, such as leaf area, height of
plant, volume of root clump, pulling resistance, etc., of 110 inbred
lines were correlated with each other and with the yield of inbred-
variety crosses. The multiple correlation between inbred-
variety yields in bushels per acre and these 12 characters of the
inbreds was .67, indicating that approximately 45 per cent of the
variability in yield of the inbred-variety crosses was dependent
upon the characters of tfye inbreds, leaving 55 per cent unac-
counted for.
208 METHODS OF PLANT BREEDING
These results show the desirability of selecting vigorous inbreds,
not only because of their value in seed production but also in
relation to the yielding ability of double crosses. It is evident
that by selecting inbred lines that have high combining ability
and making crosses between inbreds that have complementary
characters, followed by selection in self -pollinated lines, improved
inbreds can be obtained that excel both in their inherent charac-
ters and as parents of double crosses.
GENETIC DIVERSITY
Studies by Wu (1939) and Hayes and Johnson (1939) and
Johnson and Hayes (1940) of the FI crosses between these same
inbred lines have shown the value, in relation to yield of grain, of
genetic diversity of inbred lines used in double crosses. Three
groups of lines based on relationship were studied, and the yields
of single crosses were compared on the basis of origin. The three
groups may be illustrated as follows:
Inbred Cultures after Selection in Self-
Original Cross pollinated Lines
A48 X H* A94, A96
A9 X A26 A102, Alll, A116, A122, A124
A9 X A39 A99
A39 X A26 A136, A143, A145
* A48 was ail inbred from Northwestern Dent, H from Reid's, A26 from Osterland's Dent,
A39 from Rustler Dent, and A9 from Minn. 13.
Group I, no parents in common; i.e. y A94 X A102, etc.
Group II, one parent in common; i.e., A 102 X A99, etc.
Group III, both parents in common; i.e., A102 X Alll, etc.
As would be expected, group III of single crosses yielded much
less, on the average, than group I or group II, and group I of
single crosses were considerably higher yielding, on the average,
than group II.
Genetic diversity may be of as great value or of greater value
than combining ability. This is indicated by studies of Johnson
and Hayes (1940), who used inbred lines produced by the
pedigree method and made crosses only between inbred lines of
different genetic origin. These inbreds were classified into four
groups on the basis of their yields in top-crosses, in percentage of a
standard group of hybrids and Minn. 13 that was used as the
variety parent of the inbred-variety crosses. Four yield clasps
DEVELOPMENT OF METHODS OF CORN BREEDING 209
in percentage values were 80 to 89, 90 to 99, 100 to 109, and 110
or above. In the final classification, hybrids in 80 to 99 per cent
classes were considered as low combiners and 100 or above as high
combiners. The inbred lines were then studied in single crosses
in three groups of crosses: high X high, low X high, and low X
low. The single crosses were then placed in frequency distribu-
tions in comparison with recommended double crosses of like
maturity. Results are given in Table 21.
TABLE 21. SUMMARY OF FREQUENCY DISTEIBUTION OF SINGLE-CROSS
YIELDS AT THREE LOCATIONS WITH THREE REPLICATIONS AT EACH
LOCATION, WHEN COMPARED WITH RECOMMENDED DOUBLE
CROSSES OF SIMILAR MATURITY IN RELATION TO THE
COMBINING ABILITY OF THEIR INBRED PARENTS
Type of cross
Class centers of plus and minus
1 to 8 times the standard error
of a difference
Total
Mean class
-7
to
-8
-5
to
-6
-3
to
-4
-1
to
-2
+ 1
to
+2
+3
to
+4
+5
to
+6
+7
to
+8
Low X low
1
3
1
1
5
2
11
12
4
6
8
4
16
33
9
20
5
4
1
12
52
83
-0.50 0.66
+1.06 0.42
+1.10 0.24
Low X high
1
High X high
Low X low combiners yielded distinctly less in single crosses
than low X high or high X high combiners. However, low X
high and high X high yielded about the same, on the average,
when the inbred lines used in the study were genetically of rather
diverse origin. These results emphasize the value of genetic
diversity of inbred lines used in hybrid combination.
Eckhardt and Bryan (1940), at Iowa, have given data from a
series of double crosses that also emphasize the value of genetic
diversity from the standpoint of origin in relation to the yielding
ability of double crosses. If inbreds from one variety were
designated as A and B and from the other as X and F, they
compared the yields of the double cross (A X B)(X X F) with
(A X X)(B X F) or (A X F)(5 X X). The double crosses
illustrated by (A X B)(X X F) were decidedly superior to the
combination from the same inbreds (A X X)(B X F). These
facts show that crosses between inbreds that originated from
210 METHODS OF PLANT BREEDING
different varieties may be expected to be superior, on the average,
to comparable crosses from inbreds originating from the same
variety.
THE BACKCROSS
Backcross methods may be desirable in some cases to add one or
two characters to an available inbred line.
An undesirable character of the three-way cross Minhybrid 301
is smut susceptibility, dependent to a large extent on the extreme
susceptibility of inbred B164, which is used as the male parent.
The pedigree of the three-way cross is as follows:
Inbreds 11 14
\/
Fi (11 X 14) Inbred B164
3-way cross (11 X 14) X 1*164
B164 has been improved in its smut susceptibility by backcross-
ing, as illustrated in the following outline:
B164, Smut Susceptible Early Inbred 037, Smut Resistant
Method:
1. (B164 X C37) X B164.
2. [(B164 X C37) X B164] X B164.
Selection in 1 and 2 for smut-resistant plants to backcross to B164.
3. Self-pollination and selection for smut resistance for 3 years.
In field trials in 1938 and 1939 at the Waseca branch station,
several inbred lines were obtained after self-pollination and
selection had been practiced for 2 and 3 years, that during both
years had less than 10 per cent of their progeny smutted, whereas
B164 in near-by rows ranged from 85 to 90 per cent of its plants
smutted.
CONVERGENT IMPROVEMENT
This method of breeding suggested by Rickey (1927) and dis-
cussed by Richey and Sprague (1931) has been used rather exten-
sively in recent years.
The value of the method, which is equivalent to double back-
crossing, is that it furnishes a plan for the improvement of each of
two inbred lines that combine well in a single cross without
DEVELOPMENT OF METHODS OF CORN BREEDING 211
modifying the yielding ability of the single cross. Richey and
Sprague have stated that the theoretical basis for convergent
improvement assumes that:
1. Selfed lines that give a high yield in F\ carry together impor-
tant dominant genes necessary for yield and are alike in necessary
recessive genes.
2. Excess yield of the FI cross over one parent is attributable to
favorable dominant genes received from the other parent.
3. Back-pollinating a cross as N X R to R y in several successive
generations, without selection and in the absence of linkage, will
recover the genotype of the recurrent parent Rj according to the
series J^, %, %, etc.
4. Selection of the more vigorous heterozygous crosses during
the period of back-pollination will retain some of the dominant
favorable characters of N in a heterozygous condition.
5. Selection within selfed lines after back-pollination will
produce a line homozygous for R and for some dominant favorable
N genes.
6. Recovered lines N(R f ) and R(N') will differ in fewer
dominant favorable genes than N and R. Repetition of back-
crosses would gradually produce better and better lines with
more and more favorable genes in a single strain.
The more important steps in the convergent improvement
program consist of the following:
1. Selection of a high-yielding desirable FI cross.
2. Backcrossing the FI to both parents and then further back-
crossing in successive generations in two series to the respective
parents.
3. During the period of backcrossing, selection of vigorous
plants that have other desirable characters and use of these in
making the backcrosses.
4. Selection within selfed lines after several generations of
backcrossing.
5. Repetition of the four steps with the better recovered inbred
lines in order to obtain further improvement.
The practical possibilities from backcrosses have been studied
experimentally by Richey and Sprague (1931) and compared
with the theoretical as a means of learning whether heterosis is
dependent upon the interaction of dominant factors, a part of
which are contributed by each parent. In this summary, JBi, R Zj
212
METHODS OF PLANT BREEDING
Ra, etc., refer to the first, second, and third backcross progeny,
respectively. The results are summarized for six crosses. The
theoretical expectation is calculated by subtracting the yield
of the inbred R parent from the FI and assuming that one-half of
this difference will be retained in the first backcross generation,
one-fourth in the second backcross, etc., if selection is not prac-
ticed. The amount of vigor retained if selection is practiced will
represent the value of this selection. The following results were
obtained :
Cross or selfed line
Actual yields
Theoretical
(R X N)Fi
19.7 07
(R X N)Ri
11 7 + 5
11 7
(R X N)Rt
82 + 04
7 6
(R X N)R*
7.2 + 3
5.6
(R X N)R*
5.8 +03
4 6
(R X N)R 6 (only 2 crosses)
(R X N)R& (only 1 cross)
4.5 0.2
4 6 3
4.1
R selfed
3.6 + 0.2
The results give some evidence for a belief that the method
may be used to produce better inbred lines. Richey and Sprague
also studied the yields of FI crosses between the nonrecurrent
parent and the lines recovered after successive generations of
back-pollination. Except as noted, each yield given is an average
of six crosses.
Cross
(AT X R)Ft
N X (N XR 2 )*
N X (N X R,)
N X(N X R<)
N X (N X #5) (only 3 crosses)
N X (N X Rs) (only 1 cross)
(N X R)F l
*(N X R)R is written N X Rt.
Yield
9.4 0.6
13.5 0.3
15.7 0.4
17.5 0.4
18.3 0.4
17.4 0.2
17.8 0.5
Just as the yields under continuous back-pollinating methods
should approach the yield of the recurrent parent, so the yields of
crosses, between unselected back-pollinated lines in different
backcross generations and the nonrecurrent parent, should
approach the yields of FI crosses as a limit according to the series
of > etc.
DEVELOPMENT OF METHODS OF CORN BREEDING 213
It has been a common experience that backcrosses between
inbred lines of corn approach the recurrent parent in appearance
rather rapidly, and consequently it seems probable that two or
three generations of backcrossing are all that can be used and still
stand much chance of greatly changing the inbred line by the
convergent-improvement plan.
Rather extensive studies of convergent improvement have been
carried on by Hayes and Johnson (unpublished) at the Minnesota
Station. Murphy (1941) has completed one of these studies
with four inbred lines of Rustler White Dent known as C15, C16,
C19, and C20. These lines were used in two single cross com-
binations, (C15 X C19) and (C16 X C20). Within each of these
two crosses, a rather extensive convergent-improvement program
was started in 1931. In 1937, after 2 years of self-pollination and
selection, the recovered lines were crossed with the nonrecurrent
parent and the yields of these crosses compared in replicated
trials with the FI yields of (C15 X C19) or (C16 X C20). The
yields were placed in classes of plus and minus one to five times
the standard error of a difference when compared with (C15 X
C19) or (C16 X C20). The results of all crosses are summarized
in Table 22.
TABLE 22. FREQUENCY DISTRIBUTION OF YIELDS OF CROSSES FOR THE
RECOVERED LINES TESTED IN SINGLE CROSSES TO THE NONRECURRENT
PARENT. YIELDS OF RECOVERED C16 X STANDARD C20 AND
RECOVERED C/20 X STANDARD C16 ARE COMPARED WITH
STANDARD (C16 X C20). YIELDS OF RECOVERED C19 X
STANDARD C15 AND RECOVERED C15 X STANDARD
C19 ARE COMPARED WITH STANDARD
(C15 X C19)
Class centers of minus 5 to plus 2 times the
Number of
Years -back-
standard error of a difference
lines
crossed
-5
-4
-3
-2
-1
+1
+2
30
2
1
2
1
4
13
5
3
1
14
3
2
1
6
4
1
7
4
1
5
1
Of the 51 crosses tested, 1 was placed in the yield class of +2
times the standard error of a difference more than the original FI,
and 1 1 crosses were in classes of 2 to 5 of the standard error
214 METHODS OF PLANT BREEDING
of a difference and were therefore probably significantly lower in
yield than the original FI cross.
Seventeen FI crosses between recovered lines were studied in
1940. Of these 17 crosses, where both parents were recovered
lines, there were 2 crosses that were placed in the +2 class and
2 in the +4 class in comparison with the original FI crosses.
These results give sonic reason to believe that the yield of F\
crosses in themselves can be improved by the method of conver-
gent improvement. They indicate the necessity of testing the
yielding ability of recovered lines, and the results show that the
first test of a recovered line may be made by crossing with the non-
recurrent parent. All lines that do riot yield as well in crosses
to the nonrecurrent parent as the original F\ cross may be
discarded.
The improvement of an inbred line by convergent improvement
seems relatively easy for those characters in which it is seriously
lacking, and the other inbred carries these characters. In two
cases in which one of the inbred parents in a convergent-improve-
ment program was oustanding in smut resistance and lodging
resistance and gave a good yield for an inbred and in which the
other parent was deficient in these characters and gave a low
yield, it was relatively easy to improve the more undesirable
parent through convergent improvement and rather difficult to
obtain recovered lines that were superior or even equal to the
more desirable parent in yielding ability and in other important
characters.
CHAPTER XV
INHERITANCE IN MAIZE
More is known regarding the genetics of maize than of any
other organism except Drosophila. Some of the reasons why
maize has been used extensively by students of genetics may be of
interest. The plant is adapted to a wide range of environmental
conditions and shows many differential characters. It is rela-
tively easy to control pollination, and a large number of seeds can
be produced on a single ear with a single pollination. There are
many endosperm and seedling characters that can be studied in
the laboratory and greenhouse. Technics have been developed
that have made maize an especially favorable organism for cyto-
genetic studies.
Studies of the effects of self- and cross-fertilization with maize,
which started early in the present century, have furnished the
basis for the Mendelian explanation of heterosis. Later studies,
with particular attention to economic characters, including the
combining ability of inbred lines, have helped to give a genetic-
understanding of such complex characters as vigor of growth and
yielding ability and have made possible the development of
efficient breeding technics. In the short review given here, all
that will be attempted is a brief summary of those phases of
genetic studies that seem of greatest value to the student of corn
breeding.
ORIGIN AND CLASSIFICATION
A rather complete review of early theories on the origin of corn
and intensive recent studies have been made by Mangelsdorf and
Reeves (1939). A major criterion of the probable center of
origin of crop plants is the one given by Vavilov that the region of
greatest diversity of type is usually the region of origin. They
conclude that the wild ancestor of Zea mays probably occurred
somewhere in the lowlands of Paraguay, northeastern Bolivia, or
southwestern Brazil. Secondary centers of domestication include
215
216 METHODS OF PLANT BREEDING
the Andean region, Central America, and Mexico, where great
diversity of types has been observed. Mangelsdorf and Reeves
visualize maize "as a wild pod corn originating from a remote
Andropogonaceous ancestor which gave rise on the South
American continent to a single species Zea mays, on the North
American continent to a more variable genus, Tripsacum."
Maize belongs to the tribe Tripsaceae, Hitchcock (1935), and
contains three genera that are of American origin, Tripsacum,
Euchlaena, and Zea. Z. mays L. contains normally 10 pairs of
chromosomes and comprises a rather diverse group of varieties of
Indian corn. Euchlaena, called teosinte, contains two species,
E. mexicana Schrad., the annual form with 10 pairs of chromo-
somes, and E. perennis Hitchc., a perennial form of autotetraploid
type with twice the chromosome number of the annual form.
Extensive cytogenetic studies reviewed by Mangelsdorf and
Reeves lead to the conclusion that Zea and Euchlaena do not differ
widely in their chromosome make-up. In crosses between maize
and teosinte, crossing-over values, with a few exceptions, which
Mangelsdorf explains by differences in chromosome structure, are
very similar in the hybrid to values obtained in maize. It is
believed by Mangelsdorf and Reeves that teosinte was produced
by hybridization between Zea and Tripsacum. They found that
the major differences are a result primarily of four segments of
chromatin, all bearing genes with Tripsacum effects. The third
American genus Tripsacum, with n = 18 chromosomes, and Zea.
are believed by them to have descended from a remote common
ancestor. It has been found possible to hybridize Zea and
Tripsacum, and evidence indicates that they have certain genes
in common, Tripsacum differing in its evolutionary history from
Zea by a tendency to polyploidy accompanied by a perennial
habit of growth. The wild pod maize (tunicata) believed to be
the ancestor of Indian corn, presumably had its origin in South
America; Tripsacum originated in Central and North America.
It is believed that Tripsacum and Zea by hybridization and
chromosomal interchange of some sort led t6 the development of
varieties of maize that were contaminated with small additions of
chromatin from Tripsacum. Thus, maize varieties of North
America are supposed to comprise two groups: (1) pure maize,
which traces its descent to wild pod corn, and (2) Tripsacum con-
taminated maize, The evidence regarding origin of many crop
INHERITANCE IN MAIZE 217
plants is not very definite. The monograph by Mangelsdorf and
Reeves summarizes the present status of knowledge and reviews
the literature in this field. The relationships between Zea,
Euchlaena, and Tripsacum are emphasized.
Although Sturtevant (1899) divided Z. mays into several groups
and considered each to be of specific rank, several of the major
character differences are dependent upon a single factor pair. A
description of the more important groups is given here.
The Pod Corns. Each kernel is enclosed in a pod or husks; the
ear is enclosed in husks as in the other groups. The ordinary
type of pod corn is heterozygous, the homozygous type being
usually highly self-sterile. Mangelsdorf and Reeves have
described a true-breeding pod corn bearing no ears, resulting from
a combination of the homozygous condition of Tu, the pod-corn
factor, and of Ts^ the dominant factor for tassel seed. The
rachis of the ear of pod corn is definitely more brittle than in
normal maize, and there is an indication that the rachises of the
tassel are more brittle also. In the true breeding form, this
brittleness would aid in seed dissemination.
The Flint Corns. The flint corns comprise varieties with a
starchy endosperm in which the soft starch is surrounded by
corneous starch on the outside. The relative amounts of soft
and corneous starch differ widely in different varieties. Mangels-
dorf and Reeves state that the original flint corn from South
America was probably a small-seeded tropical form with large
cobs and irregular rows. Crossing with Euchlaena was believed
to produce the pointed popcorns, and backcrosses with tropical
forms led to the development of new types of flints with straight
rows.
The Popcorns. The endosperm contains only a small propor-
tion of soft starch, by far the major part of the starch-bearing
cells carrying corneous starch. There is generally some soft
starch surrounding the germ. The small size of its seeds and
cobs characterizes this group. Small, hard, pointed seeds occur
from crosses of maize with Euchlaena, and such crosses are
believed to be the original source of popcorns. Several writers
have shown that teosinte may be popped much like popcorn.
The Dent Corns. The corneous starch is located at the sides
of the seed, and the soft starch extends to the summit. The soft
starch dries more rapidly than the corneous, which causes the
218 METHODS OF PLANT BREEDING
characteristic indentation of the seed. Dent corns probably
originated in Mexico, and this is considered to be the center of
diversity of this type. Jones (1924) obtained dent types .from
hybridization of Rice popcorn and Cuzco flour corn. Wallace
and Bressman (1928) state that the dent corns of the corn belt
probably arose from a cross between a large flint type with a late
maturing type of dent that produced ears with 22 to 36 rows of
rough, very soft, shoe-peg kernels.
The Flour Corns. There is an almost complete lack of corne-
ous starch, the group being characterized by the large amount of
soft starch in the endosperm. Small amounts of corneous starch
are produced by many flour corns. The location of the small
amount of corneous starch determines whether the seed has an
indentation.
The Sweet Corns. This group is characterized by a trans-
lucent, horny appearance of the kernel and a wrinkled condition
when dry. East (1909) concluded that sweet-corn varieties are
dent, flint, or popcorns that have lost their ability to produce
starch. The few starch grains produced are small and angular.
The Waxy Corns. This group is characterized by an endo-
sperm of waxy nature resulting from a carbohydrate of different
form than in starchy varieties. The original source was China,
although waxy varieties have resulted from mutation in experi-
mental cultures.
ENDOSPERM CHARACTERS
Endosperm characters are used to differentiate several of the
major corn groups. Xenia is a result of double fertilization, the
following statement being quoted from Hayes and Garber:
Xenia may result from crossing varieties which differ in a single visible
endosperm character. When a character difference is dependent upon
a single dominant factor, xenia occurs when the factor is carried by the
male parent, or, when dominance is incomplete, xenia results when
either variety is the male. When a character difference is dependent
upon more than one factor, all located in one parent, and dominance
appears complete, xenia occurs only when these differential factors are
located in the male; when dominance is incomplete, xenia occurs if the
factors are located in either parent. When two varieties have a similar
character or a different character expression but contain between them
endosperm factors necessary for the production of a new character,
xenia occurs when either variety is the male.
INHERITANCE IN MAIZE
219
A summary of the mode of inheritance of the principal normal
endosperm characters is given in Table 23.
TABLE 23. INHERITANCE OF ENDOSPEKM CHARACTERS*
Parental type
Fi
Segregation in F*
Yellow vs. colorless
Yellow or intermediate
3 yellow:! colorless
Dominant white vs. yel-
Ivory, somewhat vari-
3 ivory:! yellow
low.
able in shade
Brown aleurone vs.
Pale yellow, partially
3 colored:! colorless
colorless.
dominant
when a single factor
pair is involved
Colored aleurone (pur-
Purple or red
Ratios 3:1, 9:7, 27:37,
ple or red) vs. color-
etc., depending on
less.
whether 1-5 factor
pairs are involved
Purple vs. red aleurone
Purple
3 purple : 1 red
Colored (purple or red)
Colorless, because of
Segregation, ratios de-
vs. colorless.
dominant inhibitory
pending on number of
factor
factor pairs involved
Starchy vs. sweet
Starchy
3 starchy : 1 sweet
Starchy vs. waxy
Starchy
3 starchy : 1 waxy
Waxy vs. sweet
Starchy
9 starchy : 3 waxy : 4
sweet
Floury vs. corneous.
No immediate effect
1 floury : 1 corneous
Normal vs. defective
Normal
Segregation 3 normal : 1
(various types of
defective when single
shrunken and shriv-
factor par is involved
eled).
* For references to literature, see Kmerson et al. (1935).
Emerson et al. (1935) have listed the genes responsible for many
of the inherited characters of maize, particularly those used in
linkage studies. Their monograph has been used freely in this
review. There are two pairs of factors for yellow endosperm
color. When both are segregating, ratios of 9 yellow: 7 white are
obtained ; when either is segregating in the presence of the homo-
zygous dominant condition of the other, ratios of 3 : 1 are obtained.
There is some evidence that genes for yellow belong to an allelic
series of various shades of yellow, although it is difficult to make
clear-cut classifications.
Hauge and Trost (1930) found a close physiological association
in dent corn between vitamin A and the yellow endosperm.
Mangelsdorf and Fraps (1931) demonstrated a direct relation
220-
METHODS OF PLANT BREEDING
between the vitamin content of corn and the number of genes for
yellow pigment in the endosperm. The average results for 2
years were as follows:
Number of genes
Factorial composition
Units of vitamin
for yellow
of endosperm
A per gram
yyy
0.05
1
yyY
2.25
2
yYY
5.00
3
YYY
7.50
-
FIG. 31. Inheritance of starchy and sweet endosperm in maize. Upper left,
ear of sweet corn with wrinkled seeds; lower left, ear of flint corn with starchy
seeds; left center, immediate result of pollinating an ear of the starchy parent with
pollen from the sweet parent; center, an Fi ear self -pollinated that segregated in a
ratio of 3 starchy: 1 sweet; upper right, a self -pollinated ear with wrinkled seeds
obtained by planting sweet seeds of the Fi. The three remaining ears at the
right were produced by self-pollinating ears produced by planting starchy seeds
of Fi plants. Note that, on the average, one out of every three ears is homozy-
gous for the starchy character. (Photograph by East.)
Brown aleurone, appearing as pale yellow, is dominant over
colorless in the absence of purple or red aleurone. There are two
factor pairs for brown aleurone, either in a dominant condition
producing color in the aleurone.
There are a series of basic pigment genes designated as 4i, 4 2 ,
A s, C, and JB, all of which are necessary for production of red
INHERITANCE IN MAIZE 221
aleurone color. When Pr also is present the color is purple.
Red or purple is epistatic to brown. When a dominant inhibitor
is present, called 7, in the presence of the basic dominant aleurone
factors for red or purple, the aleurone is colorless. In addition,
there are several intensifying or diluting factors that modify
aleurone color. There is a series of allelic factors for the R locus
and also for the / locus that cause modifications of alexirone color.
In crosses of either dent or flint corn with floury, there is no
immediate effect of double fertilization on the endosperm condi-
tion. Segregation on the ears of FI plants occurs in a 1 : 1 ratio.
Hayes and East (1915) explained these results by the hypothesis
that two genes of the floury factor are dominant over one gene for
corneous and vice versa. In crosses between dent and floury, the
floury segregate may show an indentation when there is a small
amount of corneous starch on the sides of the kernel. Classifica-
tion of floury vs. corneous is relatively easy by the use of trans-
mitted light on a ground-glass background illuminated from
below.
There are a considerable number of characters with incomplete
development of the endosperm. Most of these are lethal when
homozygous recessive and normal development of endosperm is
dominant over defective. Mangelsdorf (1926) collected 14 defec-
tives at random and made the necessary crosses to show that 13
of the 14 were due to different genetic factors. Sixteen different
defectives have now been reported. There are also at least 15
different factor pairs that are responsible for premature germina-
tion of kernels. Certain of these give 3:1 ratios when hetero-
zygous. There is one group of four duplicate factors that may
give ratios of 3:1, 15:1, 63:1, or 255:1, depending on whether
one to four pairs of factors are segregating. There are also
several pairs of factors for germless seeds. Thus, it would seem
that the development of normal endosperm is the result of the
interaction of many factor pairs.
CHLOROPHYLL VARIATIONS
There are many recessive heritable chlorophyll abnormalities in
maize. Many factors have been located in the genetic linkage
map, and it is evident that there are several factors in each
chromosome that in their interaction are responsible for the
development of chlorophyll, These rQC^ssives are pf two types,
222 METHODS OF PLANT BREEDING
those that appear in seedling progenies and those that appear in
mature plants. In a few cases, the same factor modifies chloro-
phyll development in both seedlings and mature plants.
The seedling types are frequently lethal. They include eight
or more white-seedling types, each the result of a single gene in
the homozygous recessive condition and two cases in which
duplicate genes are involved. White seedlings are devoid of
chlorophyll and generally of chloroplasts; therefore the seedlings
die when the food reserve in the seed is exhausted.
There are at least seven recessive genes for luteus seedlings.
One of these acts only in the presence of white-seedling genes;
others produce yellow seedlings in the presence of the dominant
condition of a factor for white seedlings. Most luteus types are
lethal; others give yellow seedlings and plants and therefore
produce some chlorophyll.
Twenty virescent seedling types have been described. The
seedlings are yellowish and sometimes nearly white. There is
considerable variability, ranging from types that are lethal to
those with normal development. The rapidity of turning to
green depends upon the genes involved and on temperature and
light.
There are at least 10 different genetic types of pale-green
seedlings that produce a yellowish green color in the seedling.
Some are lethal; others develop to maturity. In addition, about
37 other genes affect seedling chlorophyll color alone or both
seedlings and mature-plant color. Thus, there are at least 86
genes that affect normal chlorophyll development in the seedling.
In addition, at least 17 different genes have been described that
affect chlorophyll development in the mature plant but not in the
seedlings. The interaction of more than 100 genes is necessary,
therefore, for normal chlorophyll development.
PLANT COLOR
There are several different plant colors that are of interest to
the corn breeder. The plant and anther color resulting from the
interaction o*f several of the genes for aleurone color with genes
B and PI for plant color (Emerson et al. 1935) are given in
Table 24.
There is a series of alleles of a\ that, with other factors, affect the
development of plant, pericarp, and silk color that were given in
INHERITANCE IN MAIZE 223
TABLE 24. INTERACTIONS OP THE PLANT-COLOR GENES ai, a*, B, PI, AND R
Gene inter-
actions
With r rr
With Roff
Plant color
Anther color
Plant color
Anther color
PI
B
pi
Purple
Sun red
Purple
Pink
Purple
Sun red
Green
Green
PI
b
pi
Dilute purple
Dilute sun red
Purple
Pink
Green
Green
Green
Green
ai PI
a* B
or pi
Brown
Green
Green
Green
Brown
Green
Green
Green
a\az PI
b
pi
Green
Green
Green
Green
Green
Green
Green
Green
considerable detail by Emerson and others. These cannot be
summarized in this short review.
A series of al eles for pericarp and cob colors is of interest, P rr
is the factor for red pericarp and red cob, P for red pericarp and
white cob, P wr for white pericarp and red cob, and P ww for white
pericarp white cob. The series varies from self- or solid red
through various shades of variegation, designated as P vv .
GLOSSY SEEDLINGS
There are a number of different glossy seedlings that, in general,
have a similar phenotypic appearance that are recessive to nor-
mal. The leaves have a glossy appearance in the early seedling
stages. One of these shows the glossy character only on the third
and fourth seedling leaves, whereas the usual condition is for the
glossy appearance to show on the first seedling leaves. Classifica-
tion is made easy by sprinkling with water from a sprinkling can,
the water on glossy seedlings adhering to the leaves in large
droplets. The vigor of glossies is not greatly different from
normals, and the characters may be used to detect outcrosses in
an inbred line.
224 METHODS OF PLANT BREEDING
LINKAGE STUDIES WITH MAIZE
The genes determining the characters in maize that have been
studied fall into 10 linkage groups, corresponding to the 10 differ-
ent chromosomes. Cytological study has demonstrated that
these 10 chromosomes are morphologically distinguishable, espe-
cially at prophase in meiosis. The chromosomes are character-
ized by differences in total length, in the ratio of short to long arm
lengths, and in the position and size of terminal or subterminal
dark-staining regions. The chromosomes are numbered mainly
in order of decreasing length from 1 to 10, the number 1 being the
longest and the number 10 the shortest.
The independence of the 10 linkage groups has been established
by both cytologic and genetic studies. In addition, the linkage
groups have been identified with the particular morphological
chromosomes. Thus, the longest chromosome, 1, carries linkage
group 1, and the shortest, 10, carries linkage group 10. In all
cases, the orientation of the linkage group within the chromosome
is known, and in most cases the spindle-fiber region can be at least
approximately located in the linkage map.
In the linkage map in Fig. 32, only those genes are included
whose order is well established. The locus of the spindle fiber,
designated S.F., must be considered to be only approximate
except for group 5, genes in the part of the map above this point
being located in the shorter arm of the chromosome. The
terminal knob in chromosome 9 is shown. This map was drawn
by C. R. Burnham from information published by Emerson,
Beadle, and Fraser (1935) and from unpublished information
generously supplied by several investigators. 1 The description
of the characters was obtained from the same sources.
The location of the genes on the linkage map for each of the 10
chromosomes, with a description of the character produced, will
be given separately for each chromosome (linkage group). In all
cases, a gene symbol without subscript indicates the first or only
gene with that literal symbol.
1 Permission to use certain unpublished data in the preparation of these
linkage maps of maize was given to C. R. Burnham by L, F. Randolph,
A. C. Fraser, R. A. Emerson, M. T. Jenkins, E. W, Lindstrom, R. A. Brink,
and H. S. Perry.
INHERITANCE IN MAIZE 225
123456789 10
sr
W$ 3
cr
de Oia 2
po
Hs v, 6 0<
'knob
Rp
f:
.P^p ..
.-.S.F.
2
V92
6
bt"
11
Ig
10
12
bv
14 msg
16
og
18
d
20
in 21
C
25
27
a
35
ms, 7
30
2b 4
26
9*2
35
31
Ga
40
P p
VS 41
24
32
36
PI
v 5 24
=S.F. 28 j
roi
gl 39
sh 24
bp 38
nl
-S.F.
45
gs 2 46
In
46
Tp
43
g
49
B 4 !.
Rg
55
:=S.F. 56
sk 55
'5.F.
t8 4 56
Ts 5
51
sm 52
?j 5 1
wx
=S.F. 57
R
s:
=$
61
py
68
fl
66
sp
66
v l
74
ts 72
" i
lo U 72
v 2 71
Bn 73
w 2
r:
-S.F.
(D
e !6
86
82
br
v 4 83
na ^
zb 6
84
d 7
90
Vg
91
r
99
h
108
cm
107
Tu
111
01 112
i,
bd
120
Is
118
g! 3
*
\O "\
<=4
124
Ch ^
Cl
130
Kn
139
gs
166 bm 2
FIG. 32. Linkage map of the ten chromosomes of Zea mays, showing the
locus of the genes where the position has been determined with reasonable
certainty. The locus of the spindle fiber, designated S.F., is only approximate
in each group except group 5.
226
METHODS OF PLANT BREEDING
Chromosome 1. This is the longest chromosome, physically,
of the 10. The allelic series of P, which produces pericarp and
cob colors, is located in this chromosome. The gene order and
locus of 15 genes have been determined. These are listed below:
0-sr Striate. Leaves with fine longitudinal striations throughout the
life of the plant.
25-wsn Male sterile- 17. Anthers usually not exserted. Some pollen
occasionally shed.
27-/S2 Tassel seed-2. Terminal inflorescence usually completely pis-
tillate; no pollen produced. Ear develops if terminal inflores-
cence is removed soon after emergence. Secondary florets in
ear development, giving irregular arrangement of kernels.
Tassel seed-2 is similar to ts (chromosome 2) except that the
plants are usually stronger, and terminal inflorescence is less
compact.
28-P Pericarp and cob color. A large series of allelic genes for pericarp
and cob color.
30-zZ Zygotic lethal. Lethal that kills the very young sporophyte and
endosperm.
86-264 Zebra striped-4. Seedlings with irregular chlorotic crossbands on
leaves in early stages. Bands may disappear as plants get
older.
55-as Asynaptic. Partially sterile type characterized by lack of associ-
ation of homologous chromosomes during the first meiotic
division. Produces a few kernels when pollinated by normal
plants. Usually sheds no pollen.
86-frr Brachytic. Culm notably shortened as a result of shortening of
internodes. Plants one-fourth to one-half normal height.
Leaves stiff and straight.
90-F(7 Vestigial glume. Glumes of tassel and flowering glumes of cobs
greatly reduced. Anthers of tassel exposed, pollen shed only
occasionally.
91-/ Fine stripe. Seedling virescent. Plant shows fine stripes of
white tissue in the leaf blades, only rarely in the auricles.
108-an Anther ear. Leaves broad. Plant variable in size, often almost
normal in stature. Stamens develop throughout pistillate
inflorescence. Ear usually ends in an unbranched spike con-
sisting of staminatS flowers only.
120-7^3 Tassel seed-3. Similar to is and ts z except that the terminal
inflorescence usually is mixed pistillate and staminate. Usually
pollen can be obtained. Secondary florets develop in ears.
130- Kn Knotted leaf. Overgrowth of areas of midrib and other vascular
tissue resulting in kinking or knotting of the veins.
139-0S Green striped. Leaves in three- or four-leaf stage, and later,
show light green stripes between main vascular bundles.
Plant weak.
INHERITANCE IN MAIZE 227
166-&W2 Brown midrib-2. Brown color develops in midrib and over
vascular bundles of leaf blade and sheath. Similar to bm but
less intense.
Chromosome 2. The factor pair for flinty vs. floury endosperm
(Ff) is located in this chromosome. The location of 10 genes in
this chromosome are:
White sheath-3. Partial absence of chlorophyll in culms and
sheaths of seedling and older plants.
ll-lg Liguleless leaf. Leaf usually lacks ligule and auricles and stands
upright at base.
Glossy seedling-2. Seedling character. Younger leaves have
glossy appearance, visible in bright sunlight.
Green striped-2. Mature plant with green stripes.
49-B Booster. Plant-color intensifier. In appropriate genotypes, gives
intense sun-red, purple, or brown plant color.
56-sfc Silkless. Pistils abort. No silks. Plants female sterile. Cobs
grow normally and contain many anthers.
68-// Floury endosperm. Endosperm floury (noncorneous). Female
contribution to endosperm determines character; fl fl Fl gives
floury endosperm, and Fl Flfl gives normal (flinty) endosperm.
74-ts Tassel seed. Terminal inflorescence usually completely pistillate,
no pollen produced. Ears develop if terminal inflorescence is
removed soon after emergence.
82-i>4 Virescent seedling-4. Seedlings yellowish green. Plants turn
green slowly and may be distinguished from normal plants later
than can most virescent seedlings.
124-CVi Chocolate pericarp. Pericarp dark brown or chocolate in color.
Chromosome 3. The allelic series A, A b , a p , a is in this
chromosome. These genes are essential to the development of
plant, aleurone, and pericarp colors. The 11 genes in this
chromosome are:
0-cr Crinkly leaf. Plants somewhat shorter than normal. Leaves
broad, with characteristic crinkling at base. *
18-d Dwarf plant. Plant of very low stature, with broad thick leaves.
Starninate inflorescence compacted. Stamens produced in ears.
26-m 2 Ramose ear-2. Much less extreme than ra in tassel and ear.
46-L0s Liguleless leaf-3. Only a portion of the ligule present.
4&-Rg Ragged leaf. Chlorotic areas in leaves of older plants, leaves
becoming much split and torn. Character shows when plants are
about half-grown.
55-fe* Tassel seed-4. Terminal inflorescence produces staminate and
pistillate flowers. Usually few kernels produced in the tassel.
Secondary florets develop in ears. Pollen usually shed.
228 METHODS OF PLANT BREEDING
72-6a Barren stalk. Plant characterized by absence of pistillate inflores-
cence. Stem circular in cross section, lacking the characteristic
groove.
83-na Nana. Plants dwarfed, from one-fourth to one-third normal height.
- Leaves characteristically stiff and twisted.
111-a Anthocyanin. Plant, aleurone, and pericarp color. Inappropriate
genotypes, gives green or brown plants, colorless aleurone and
brown pericarp.
123-etf Etched endosperm. Endosperm scarred, seedling virescent.
Chromosome 4. The factor pair for starchy vs. sugary endo-
sperm (Su su) is located in this chromosome. The factor produc-
ing tunicate plants (Tu) (pod corn) is also in this chromosome.
The order of 11 genes has been determined. The exact location
of the spindle fiber has not been determined. It is near silkless
(sk) and is indicated on Fig. 32 by dotted lines.
Q-de Defective endosperm. Incomplete development of the endosperm.
Viability poor.
35-Oa Gametophyte factor. Ga pollen, in competition with ga pollen on
Ga silks, functions in the production of 95 to 99 per cent of the
kernels.
Tassel seed-5. Tassel contains both silks and anthers and is not
compacted. Usually few kernels develop in tassel. Secondary
florets develop in ears.
Small pollen. Pollen grains small but filled with starch. Trans-
mitted usually through the ovules only.
71-su Sugary endosperm. Endosperm translucent and wrinkled.
73-/o Lethal ovule. Ovules abort. Gene transmitted almost wholly by
pollen.
75-cfeie Defective endosperm- 16. Incomplete development of endosperm.
Lethal.
84-Z&0 Zebra striped-6. Chlorotic crossbands in leaves of nearly mature
plants.
107- Tu Tunicate ear. Glumes in both staminate and pistillate inflores-
cence long, enclosing individual kernels in 'ear more or less
completely.
112-ja Japonica-2. Variegated striping. Expressed in seedlings as well
as mature plants, some seedlings nearly white.
118-gZs Glossy seedling-3. Glossy surface on younger leaves.
Chromosome 5. The factor pair for purple vs. red aleurone
(Pr pr) is in this chromosome. The location of eight genes and
the characters produced by them is as follows:
0-a2 Anthocyanin-2. Dominant allele complementary to the Aa pair in
the production of plant and aleurone colors. Has no effect upon
pericarp color.
INHERITANCE 1$ MAIZE
Brown midrib. Brown color develops in midrib and over vascular
bundles of leaf blade and sheath. Character appears in three- to
four-leaf stage but shows better at later stages.
S.F. Spindle fiber. Known to be between bm and bt.
8-bt Brittle endosperm. Endosperm translucent, usually shrunken and
wrinkled.
10-Vs Virescent seedling-3. Seedling light yellow but turns green quickly.
12-bv Brevis. Plants usually about one-half normal height, owing to
shortening of internodes in region of pistillate inflorescence.
Bl-pr Red alcurone. In presence of other genes necessary for aleurone
and scutellum color, gives red aleurone and scutellum as con-
trasted with purple in presence of Pr.
40-?/s Yellow stripe. Leaves show yellow stripes between main vascular
bundles.
72-Vz Virescent seedling-2. Seedlings very light yellow. Plants turn
green rather slowly.
Chromosome 6. This chromosome carries the factor pair for
yellow vs. white endosperm (Yy) and the plant-color factor pair
(PI pi). Five genes have been definitely located as follows:
0-pc Polymitotic. Plants partially sterile. Young microsphore cells
undergo several mitotic-like divisions in rapid succession without
division of the chromosomes. No pollen shed; few seeds pro-
duced in crosses with normal.
13-F Yellow endosperm.
4:l-Pl Purple plant color. In appropriate genotypes gives dilute purple,
intense purple, or brown plants.
51-sm Salmon silk. In presence of red pericarp (P rr , etc.) silks are salmon
in color. In absence of pericarp color, silks are brown.
61-p?/ Pygmy. Plant short, with short, thick, striated leaves.
Chromosome 7. The gene for brown aleurone (Bri) is located
in this chromosome. Nine genes have been located in this
chromosome. These are:
Q-Hs Hairy sheath. Leaf sheaths hairy throughout development.
20-in Intensifier of aleurone color. Intensifies purple and red aleurone.
24~z>6 Virescent seedling-5. Seedlings greenish yellow, turn green very
quickly.
32-ra Ramose ear. Ear much branched throughout, conical. Tassel
much branched, conical in shape.
36-grZ Glossy seedling. Leaves have glossy appearance.
46- Tp Teopod. Plant strongly tillered, with narrow leaves. Number of
nodes greater than in normal plants. Many small podded ears.
Staminate inflorescence with long bracts, many plants not
shedding pollen.
52-tj lojap striping. Variegated stripe tnat shows throughout life of
the plant. Varies from albino to variegated.
230 MJSTtiuDS OF PLANT BREEDING
71-Bn Brown aleurone. Pale yellowish aleurone color. Shows only in
absence of purple and red aleurone. Often confused with light
yellow endosperm.
109-&d Branched silkless. Ears branched at base, often without silks.
Tassel has characteristic branches, the spikelets occurring in
groups of more than two.
Chromosome 8. Fewer genes have been located in this
chromosome than in any other. Of the three genes known to be
in this chromosome, all are in the long arm. Consequently, the
spindle fiber is not shown in Fig. 32. The order of the known
genes is as follows:
O-^ie Virescent seedling-16. Seedlings yellowish green.
I4-mss Male sterile-8. No anthers exserted. Microsporocytes usually
disintegrate.
28-j Japonica. Variegated striping in leaves and sheath. Does not
show in seedling stage.
Chromosome 9. One of the basic aleurone color factors (C) is
located in this chromosome. So is the gene for waxy endosperm
(wx) . This chromosome, in certain stocks, has a terminal knob at
the end of the short arm. Six genes have been placed in order on
this chromosome, five being in the short arm. These are as
follows :
0-knob Terminal knob on the chromosome.
2-ygz Yellow green-2. Seedling and plant yellowish.
"21-C Aleurone color. In appropriate genotypes, gives purple or red
aleurone.
24-sft Shrunken endosperm. Endosperm shrinks during drying stage at
maturity, giving a smooth indentation at the crown or a collapse
at the sides of the kernel.
39-6p Brown pericarp. In presence of P, gives brown pericarp.
54-wx Waxy endosperm. Waxy starch in endosperm; embryo sac and
pollen grains stain reddish brown with iodine solution, as con-
trasted with normal starch, which stains blue.
66-u Virescent seedling- 1. Seedlings yellowish, become green relatively
early in development.
Chromosome 10. This is the shortest chromosome, in terms of
physical length, of the 10. It has a genetic map length of 99
units. One of the basic aleurone and plant-color factor pairs (Rr)
is located in this chromosome. The locations of 8 genes have
been mapped.
INHERITANCE IN MAIZE 231
Resistance to leaf rust. Resistance to physiologic race 3 of
Puccinia sorghi.
16-00 Old gold. Dominant chlorophyll striping. Light-green or yellow
striping begins after 5- 6-leaf stage.
24-nZ Narrow leaf. Plants weaker than normal, with narrow leaf blades.
Leaves tend to be longitudinally striated, like lineate (li).
38-Z 8 Luteus-8.
43-0 Golden. Full-grown plants of a yellowish green color.
57-JK Colored aleurone and plant. In appropriate genotypes, gives purple
or red aleurone. Exists in a series of alleles affecting aleurone,
plant, and anther color.
73-w 2 White seedling-2. Seedling albino, devoid of chlorophyll.
84-dr Dwarf-7. Plant of low stature.
99- 2 Luteus-2. Yellow seedling.
In addition to the 86 genes whose location on the chromosome
map has been determined with reasonable certainty, there are
about 108 other genes that have been placed in particular
chromosomes, although the location on the map, in relation to the
genes whose locus is known, has not been determined. Some 102
chromosome translocations have been found in which the chromo-
somes involved have been determined.
INHERITANCE OF QUANTITATIVE CHARACTERS
Studies on the inheritance of quantitative characters in maize
were started by East, in 1906, in Connecticut, and a little later by
Emerson, in Nebraska. These and other experiments created a
wide interest in the multiple-factor explanation of the inheritance
of size characters. It is rather generally accepted that many
normal characters are the result of the interaction of many genetic
factors. A method commonly used with size characters is to
cross parents that differ rather widely in a character, such as
length of ear in maize, and study the Fi, F^ and F 3 generations in
comparison with the parents.
For quantitative characters, dominance is often incomplete or
lacking. When dominance is complete, the expected ratios may
be obtained by the expansion of the binomial (3 + l) n , where n
is the number of allelic pairs of factors. When the hetero-
zygous condition of a factor pair gives half the effect of the
dominant homozygous condition and there is a cumulative effect
of one factor on another and all factor pairs are of equal value in
their effect on the character, the expected ratios in F^ may be
obtained by the expansion of the binomial (1 + l) 2n . Where n
232
METHODS OF PLANT BREEDING
is 3, for example, or three factor pairs are involved, the expected
ratio will be 1:6:15:20:15:6:1.
Such a ratio approaches the normal curve, and when suffi-
cient Fz individuals are studied the parental combination of
characters should be recovered. If each of the parents contains
different factors that have an effect on the character, illustrated
by the cross of aaBB X AAbbj types will be obtained in F% and
later generations that exceed the limit of the parents. In actual
practice, there is no reason to expect that all factors have like
value in their effect on the character. This will affect the form
of the curve but not its regularity in the absence of dominance.
With partial to complete dominance, the curve will be of the skew
types but cannot easily be distinguished from normal when a large
number of factor pairs are segregating.
An illustration of the usual type of data that are obtained,
where dominance is incomplete, may be observed from a cross
between Tom Thumb pop with Black Mexican sweet, as given by
Emerson and East (1913).
TABLE 25. FREQUENCY DISTRIBUTION FOR LENGTH OF EARS IN THE
PARENTS Fi, F 2l AND F B GENERATIONS OF A CROSS BETWEEN TOM
THUMB POP AND BLACK MEXICAN SWEET CORN
Ear classes, cm
Gen-
Parent or
Parent
cross
class
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Mean
Tom Thumb
p
/\
1
94
s
6 6 07
Black Mexi-
P
3
11
12
15
26
15
10
7
2
16.8 .12
can
Fi
1
\?.
1?
14
17
q
4
12.1 -1- .12
F 2
4
5
22
56
80
145
129
91
63
27
17
6
1
12.7 .06
Ft
11
1
5
8
9
25
20
2
12.7 .11
F 8
18
9
7
13
18
25
20
6
4
15.8 .13
Fa
10
1
8
22
25
15
22
7
2
10.5 .10
F s
9
5
13
20
28
28
30
10
7
6
3
2
1
10.0 .12
Ft
10
3
10
33
47
40
18
3
9.2 .07
From these results, it will be noted that the Fi was intermediate
in ear length but no more variable than the parents and that the
Fz was more variable than the parents. The few F% lines illus-
trated give wide differences; the shortest eared line had a mean
of 9.2 .07 and the longest, 15.8 .13. If sufficient lines had
been tested, it is reasonable to expect that the parental forms
INHERITANCE IN MAIZE
233
would have been recovered, the frequency of such recovery
depending upon the number of factor pairs involved and the
nature of their interaction.
Similar data were given by Emerson and East for diameter of
ears, weight of seeds, breadth of seeds, height of plants, number of
nodes per stalk, internode length, number of stalks per plant,
total length of stalks per plant, and duration of growth. For
these quantitative characters, it should be emphasized that when
several pairs of factors are involved it may be difficult to recover
the parental types in JFV The usual method used by the breeder
is to grow as large a population in F 2 as can be studied and select
for the character desired. Recovery of the parental types may
be obtained, as a rule, by continuing selection in F 3 and in the
later segregating generations.
LINKAGE OF FACTORS FOR ROW NUMBER WITH GENES AT
KNOWN LOCI
Further proof of the multiple-factor explanation of the inherit-
ance of quantitative characters has been obtained by studies of
TABLE 26. COB COLOR AND Row NUMBER IN THE F 2 OF THE CROSS OF
IODENT X GOLDEN BANTAM AND IN THE BACKCROSS
Rows
Fi generation
FI X Golden Bantam
per ear
R*
r
72
r
8
24
62
10
3
4
94
109
12
40
17
116
101
14
46
6
19
10
16
17
3
3
18
2
Total
108
30
256
282
P
0.03
0.0001
* In this table, R stands for red cob arid r for white.
linkage between quantitative characters with genes that are
known to be located in particular loci of the chromosome map.
Studies have been made by Lindstrom (1931) of the linkage
relations of row number with the gene P for pericarp and cob
color, the R factor for aleurone color, Su for starchy endosperm.
234 METHODS OF PLANT BREEDING
and Y for yellow endosperm color. An illustration of the type of
results obtained will be given for a cross of lodent with a modal
value of 16 rows per ear and red cob X Golden Bantam, an
8-rowed, white-cobbed variety. The extent of association was
determined by the calculation of X 2 and P for independence.
In both F% and the backcross, it is evident that the white-
cobbed (r) ears average lower in row number than the red-cobbed
ears. The data given illustrate one of a rather extensive series of
crosses where there appeared to be an association between row
number and cob color. This linkage relation seems best explained
by the hypothesis that one of the factor pairs for row number is
located on chromosome 1 and shows genetic linkage with one of
the allelic pairs of factors for pericarp and cob color.
In studies of linkage relations between row number and the
factor pair Su su for starchy-sugary endosperm, crosses were
made in both phases, i.e., high row, sugary X low row, starchy
and low row, sugary X high row, starchy. In certain crosses of
both types, there was definite evidence of genetic linkage; other
crosses did not show association.
With the Yy factor pair, there was some evidence of a loose
linkage. The greater part of the data also showed evidence of
linkage between the factor pair for aleurorie color Rr and row
number.
INHERITANCE OF SMUT REACTION
Studies of linkages for smut reaction have been made by vari-
ous workers. Immer (1927) and Hoover (1932) made crosses
between resistant inbreds and genetic testers that were suscepti-
ble. Most of the cases of association were between characters
such as tassel seed, brachytic and liguleless, and susceptibility.
These are of such a nature that the association may be explained
on the basis that the morphological character tends to make the
plant more susceptible. Immer observed an association between
the pericarp factor pair Pp and smut reaction; Hoover obtained
evidence in certain crosses, but not in others, of linkages between
smut reaction and su } v%, and sh wx located in chromosomes 4, 5,
and 9, respectively.
More recently, studies of linkage of smut reaction have been
made by the use of chromosome interchanges. In these experi-
ments, smut-resistant inbred lines were crossed with particular
INHERITANCE IN MAIZE
235
interchanges that were smut-susceptible. In the studies of
Burnham and Cartledge (1939), the FI was rather highly resistant,
and this was outcrossed to a normal susceptible inbred. Inter-
change plants can be differentiated from normal, since they
produce approximately 50 per cent aborted pollen grains. In the
experiments of Saboe and Hayes (1941), the Fi cross between a
resistant inbred and a susceptible interchange was intermediate in
susceptibility. The Fi was backcrossed to the resistant inbred
parent.
In both experiments, the plants in the segregating families were
first classed as normal or interchange on the basis of pollen
sterility, and data were taken on smut reaction. A portion of
Burnham and Cartledge's results will be summarized to show how
linkages were determined.
TABLE 27. REACTION TO SMUT IN THE PROGENY OF BACKCBOSSES OF
THE Fi (RESISTANT INBEED X SUSCEPTIBLE INTERCHANGE) X SUS-
CEPTIBLE NORMAL
Normal
Semisterile
^1
P
cross
Smutted
Not smutted
Smutted
Not smutted
l-2a X resistant
57
215
55
198
0.80
l-2c X resistant. .
53
261
91
257
<0.01
3-8a X resistant
210
602
250
488
<0.01
l-9c X resistant.
10
38
17
34
0.20
Highly significant deviations from random expectation were
obtained with interchanges involving l-2c and 3-8a, as shown in
the table, and also with l-6a, l-9t>, 2-6a, and 6~8a. In the case
of the interchange l-2c, where definite linkage is noted, and the
point of interchange is close to l-Qc in chromosome 1, where there
was no evidence of linkage, it seems probable that the linkage
relation is with chromosome 2. Break l~2c occurred near the
location v in the longer arm of chromosome 2.
In the studies by Saboe and Hayes, significant associations
were observed with interchanges 3-76, 5-7dI, 6-9a, and 8-10a in
the crosses of interchanges with the resistant inbred line of Minn.
13 and with interchanges l~4a, 3-5c> and 5-8a in crosses with the
resistant inbred from Rustler.
236 METHODS OF PLANT BREEDING
From these results, it is evident that there are many loci for
reaction to smut. Various investigators, including Jones (1920),
Hayes et al. (1924), and Garber and Quisenberry (1925), have
found it relatively easy to isolate resistant inbred lines by selec-
tion in self-fertilized lines. It is possible that the number of
factor pairs for smut resistance is not necessarily very great from
any one source of origin.
INHERITANCE OF COMBINING ABILITY
There is a great deal of information that leads to the conclusion
that some inbred lines combine well in top-crosses or with most
unrelated lines, whereas other inbreds rather generally have
lower combining ability. This entire problem has received some
consideration in the chapter on Breeding Methods, but some of
the more salient facts in relation to inheritance of combining
ability will be reviewed briefly. Experiments at Minnesota and
Iowa, already reviewed, show that there is a direct correlation
between the vigor of inbred lines and their yielding ability in top-
crosses. It is equally evident that the relationship is not easily
measured by the eye, since it is impossible to determine by inspec-
tion whether a particular inbred will give high or low yields, on
the average, in crosses.
There is general acceptance that the yielding ability of inbreds,
as determined by their crosses, is dependent upon the number and
nature of dominant growth factors of each inbred in relation to
the dominant factors carried by the other parent. This has led
to the test of inbred lines for their combining ability and the
selection of genetically diverse lines to use in any particular
hybrid. Davis (1929) first suggested the use of top-crosses to
test combining ability, although the general wide use of the
method should be credited to the work of Jenkins and Brunson
(1932). Wu (1939), Hayes and Johnson (1939), and Johnson
and Hayes (1940) have summarized extensive studies that show
that yielding ability in single crosses is greater, on the average, in
lines that are unrelated on the basis of origin than in lines of a
somewhat similar origin. Using inbreds of diverse origin that
had been classified on the basis of combining ability in top-crosses
into two groups for yielding ability, low and high, Johnson and
Hayes (1940) found that, on the average, crosses between low X
low yielded less than low X high or high X high, although F\
INHERITANCE IN MAIZE
237
crosses between low X high yielded as well, on the average, as F\
crosses between high X high*
Eckhardt and Bryan (1940) selected inbreds from two different
varieties and called those from one variety A and B and from the
other, X and Y. From any double cross, where two inbreds were
selected from one variety and two from the other, the yield of
(A X B) X (X X F) was significantly greater than (A X X) X
(B X F) or (A X F) X (B X X).
Several studies on the inheritance of combining ability have
been made. Jenkins (1935) concluded that inbred lines showed
their individuality as parents in top-crosses in the early segregat-
ing generations and remained relatively stable in later inbred
generations. This has led to a consideration of the value of test-
ing combining ability in the early generations of selfing and
continuing selection in self-pollinated lines from lines of high
combining ability. Jenkins explained these results on the basis
that combining ability was controlled * by a large number
of dominant genes and that the effect of different genes was of
approximately the same value. He thought equal numbers of
favorable dominant genes would be preserved by chance through
the successive generations of selfing.
In a recent study, Jenkins (1940) selected seven inbred lines of
the variety Krug that had been tested in top-crosses for each of 4
successive years, the tests having been begun after the lines had
been selfed for 3 years. The results of these trials are as follows:
Inbred
Acre yield in top-crosses
1930
1931
1932
1933
Mean
K679
39.7
37.3
51.9
22.4
37.9
31.1
36.2
37.5
75.8
74.6
81.1
70.4
79.8
79.5
76.4
76.5
71.3
81.8
77.6
74.7
79.5
66.1
71.7
75.1
80.9
92.3
84.3
79 2
86.3
73.6
82.0
79.6
66.9
71.5
73.7
61.7
70.9
62.6
66.6
67.3
K682.
K683
K685
K686
K687
K689
Krug variety
Remnant seed of the first-year selfs Si was used for each line,
and within each line pollen of each of 16 plants was applied to the
238
METHODS OF PLANT BREEDING
silks of 25 plants of the Krug variety. Seed for each top-cross
was obtained by mixing the seed of 25 ears, and the 112 top-
crosses so made were tested in replicated yield trials. The
analysis of variance is as follows :
TABLF- 28 ANALYSIS OF VARIANCE OF YIELDS OF TOP-CROSSES OF INDI-
VIDUAL PLANTS IN ONE-GENERATION KRUG SELFS
Source of variation
Degrees of
freedom
Mean
squares
F
Lines
6
680 32
34 07*
Sibling plants within lines
105
77 21
3.87*
Replications within lines
63
403 45
20.20*
Error
945
19 97
Total
1119
* Highly significant.
On the basis that heterozygosity, and therefore variance, will be
reduced among siblings within lines in the various succeeding
generations of selfing according to the series %, ^ %, etc.,
Jenkins determined the probable significance of segregation for
combining ability. The variance for sibling plants within lines
in the first selfed generation was 77.21. This is composed of error
mean square 19.97 and variance due to genetic differences or
57.24. The calculated mean squares for sibling plants in the
various generations of selfing from Si to $ 8 were Si 77.21, $ 2
48.59, S s 34.28, S 4 27.13, S, 23.55, S 6 21.76, S 7 20.86, and S 8 20.42,
respectively. Although all generations except $ 8 showed highly
significant calculated variations, it is apparent that the chances
for segregation were much the greatest in the early generations of
selfing and on the average became progressively less as selection
progressed.
Sprague and Bryan (1941) studied segregation for yield, lodging,
and damaged kernels in top-crosses of 73 F% lines selected from a
single cross between inbreds. Twelve F 3 lines were selected that
represented low, medium, and high yielders in top-crosses and
that differed in damaged kernels and lodging resistance. Five F&
plants were chosen from each F* line and top-crossed to the
synthetic hybrid 8037 and tested in yield trials for 1938 and 1939.
Highly significant variances between F 4 lines within F 3 families
INHERITANCE IN MAIZE 239
were obtained for yield and lodging, and a significant variance
on the basis of odds of 20 : 1 was obtained for damaged kernels.
Hayes and Johnson (1939) studied 110 inbreds obtained from
selection through F G in the progeny of single crosses between
inbred lines. These crosses comprised three general types of
crosses between inbreds that were classified as low and high
combiners in top-crosses. From crosses between low X low,
most of the inbreds selected proved low in combining ability.
Crosses of low X high combiners and subsequent selection
through FG gave both low and high combining lines, whereas from
crosses of high X high and subsequent selection in selfed lines
through F& only high combining lines were isolated.
INHERITANCE OF OTHER IMPORTANT CHARACTERS
A complete review of important studies of inheritance in maize
cannot be made in the space available. Inheritance of protein
content has been studied by Hayes and Garber (1919), East and
Jones (1921), and Hayes (1922). Crosses between low and high
protein lines have low protein content in F\. It seems probable
that many genes are responsible for the inheritance of protein.
East and Jones concluded that the apparent dominance of low
protein over high protein was a result of heterosis, which is a
further indication of the multiple-factor theory of protein
inheritance.
Jenkins (1932) has shown that inbred lines and crosses in corn
have differential resistance to heat and drought. Haber (1938)
obtained similar results with inbred strains of sweet corn. Heyne
and Brunson (1940) have found also that selfed lines of corn can
be isolated that differ in reaction to heat and drought. Heat
tolerance was definitely inherited and usually intermediate to
dominant in FI. A case of linkage of reaction to heat with the
Pr pr factor pair was noted. The su gene was directly responsible
for susceptibility to heat while certain of the glossy seedling genes
gl and gl% apparently protected the seedlings from heat.
Holbert and Burlison (1928) noted marked differences in reac-
tion to cold between inbred lines and within commercial varieties.
Most maize plants are proterandrous, the pollen being shed 3
to 5 days before the silks appeal*. A variety of popcorn from
Spain was found to have proterogynous habit, the pollen being
shed 2 or 3 days after the silks emerge, which is the normal condi-
240 METHODS OF PLANT BREEDING
tion in Tripsacum and Euchlaena. The inheritance of this
character has been studied by Kempton (1924). The proter-
ogynous strain used in crosses averaged 2.96 0.18 days from
silking to pollen shedding; the proterandrous strain shed pollen
2.3 0.11 days before the silks appeared. No proterandrous
plants were found in the proterogynous strain. The proter-
ogynous strain produced an occasional plant that showed a
tendency to be proterandrous. The proterogynous strain also
produced several plants that failed to extrude anthers and never
shed pollen. In crosses between the two strains, the F\ was
proterandrous. Segregation occurred in Fz, the number of
proterogynous plants obtained being too few for a simple
Mendelian ratio. Male sterile plants appeared also, and the
conclusion was reached that proterogyny was a result of a variable
expression of the male sterile condition brought about by modify-
ing factors.
Pericarp tenderness has been found by Johnson and Hayes
(1938) to be an inherited character. Different inbred lines give
wide deviations for the mean expression of tenderness. The
number of factor pairs involved was not determined, but the
results proved that it was relatively easy to modify the tenderness
of an inbred line of sweet corn by a process of crossing, backcross-
ing, and .selection.
Harvey (1939) has reviewed previous studies of differential
responses of corn to various levels of fertility. The various
studies show clearly that inbred lines and their FI hybrids fre-
quently show differential response to various nutrients, including
phosphorus and nitrogen, and in water economy. Harvey dealt
with the absorption and utilization of nitrogen ionic forms by
corn inbreds and hybrids. The inbred strains and their F\
hybrids were grown in aqueous mineral solutions. Differential
response to ammonium and nitrate nitrogen was statistically
significant. Some strains made relatively more growth than
other strains on ammonium nitrogen compared with their growth
on nitrate nitrogen. The response of FI crosses indicated that
there was a partial dominance of the genetic complex for efficient
utilization of ammonium nitrogen.
There are wide differences among varieties, inbred lines, and
hybrids in reaction to important diseases and insect pests. Mains
(1931) studied reaction to leaf rust, Puccinia sorghi, using physi-
INHERITANCE IN MAIZE 241
ologic races 1 and 3. Resistance to both races was due to the
same genetic factor. In crosses of resistant X susceptible,
resistance was dominant, and segregation in F 2 was on the basis
of 3 resistant: 1 susceptible. Ivanoff and Riker (1936) and
Wellhausen (1937) have shown that resistance to bacterial wilt
was an inherited character. In general, in crosses of resistant
with susceptible inbreds, resistance behaves as a dominant.
Wellhausen concluded that there were at least three pairs of
factors, independently inherited, that condition resistance. The
presence of all three dominant factors in either a heterozygous or
a homozygous condition resulted in a high degree of resistance.
Differences in reaction to ear, stalk, and root rots have been
observed by many investigators. The mode of inheritance has
not been worked out in detail.
Resistance to insect pests has been studied by several workers.
The leaf aphid, Aphis maidis Fitch, attacks the strains of suscepti-
ble plants and prevents pollen shedding. Snelling et al. (1940)
reviewed the literature on resistance to aphis attack and presented
data on inbreds and crosses between them to show that resistance
to aphis injury is a heritable character. There have been
numerous studies of resistance to the European corn borer.
Marston (1930) studied crosses of Maize Amargo, a resistant
variety, with Michigan varieties and concluded that reaction to
the borer segregated in a ratio of 3 susceptible: 1 resistant.
Meyers et al. (1937) found resistance to be an inherited character
but conclude " no thing suggestive of immunity nor of a genetically
simple resistance was found." Inherited resistance to the corn-
ear worm was reported many years ago by Collins and Kempton
(1917), and considerable progress has been made in the develop-
ment of resistant varieties. Blanchard et al. (1941) found that
some inbiled lines were resistant to the corn-ear worm, whereas
others were susceptible. Some resistant inbreds transmit a high
degree of resistance to their FI crosses with either resistant or
susceptible inbreds. Other FI crosses of resistant X susceptible
did not show a dominance of resistance. FI crosses of susceptible
inbreds were generally susceptible, although one case of a
resistant FI was obtained from a cross of susceptible inbreds.
Marked differences in reaction to the chinch bug have been noted
by Snelling and Dahms (1937).
CHAPTER XVI
CONTROLLED POLLINATION METHODS OF BREEDING
CROSS-POLLINATED PLANTS
Darwin (1876) made the first carefully controlled extensive
experiments of the effects of self-fertilization. He noted the
great uniformity of inbred lines and in general a marked reductior
in vigor in self -pollinated lines, although he recorded exceptions
In several cases he found little harmful effect of continued self-
fertilization after the first generation. He observed that con-
tinued brother-sister mating had the same effect as continued
self-fertilization. He believed, however, that this similarity wag
the result of growing the inbred cultures under the same environ-
mental conditions, for he found that crosses between his inbred
stocks and those from another locality were very vigorous.
In spite of these results, Darwin agreed with Knight that self-
fertilization was not a natural process. They were the chiei
exponents of the so-called Knight-Darwin law that " nature
abhors perpetual self-fertilization.^ The vigor of FI crosses wa&
explained by Darwin on the basis of germinal differences con-
tributed by the parents.
EFFECTS OF SELF-FERTILIZATION
East and Jones (1919) summarized many of the experiments on
the effects of inbreeding and on hybrid vigor and gave what seems
to be a sound biological explanation of the results. This mono-
graph furnishes a wealth of information for the student of plant
breeding.
In its application to plant and animal improvement, inbreeding
gives an opportunity for controlled selection and in this way aids
in the rapid isolation of strains homozygous for the desired char-
acters. In many cases, the vigor of growth of a plant or animal
is dependent upon the interaction of a large number of growth
factors. Most of these factors are dominant or partially domi-
242
CONTROLLED POLLINATION METHODS 243
nant in hybrids, and, because of their number, linkage is involved.
Inbreeding tends to reduce the number of heterozygous pairs of
growth factors present in the inbred line of the organism.
In self -pollinated crops, natural and artificial selection has led
to the development of vigorous inbred lines. It would seem that
artificial inbreeding and selection with cross-pollinated crops
might be expected to accomplish similar results. Continued
studies of the effects of inbreeding and selection with cross-
pollinated crops plants show the value of the methods, although
there are many instances where the reduction in vigor is so great
that inbreeding cannot be continued for many generations with
the hope of obtaining inbred lines that are vigorous.
Studies of self-fertilization with corn and other organisms, since
the rediscovery of Mendel's laws, have furnished the basis for the
Mendelian explanation of hybrid vigor and the partial standard-
ization of breeding technic with cross-pollinated plants. The
extent to which controlled inbreeding can be used, the desirability
of breeding by adding the factor for self-fertility to many lines
when it is available in the organism and where extensive self-
sterility is involved, as well as many other similar problems, can
be solved only by extensive study with each particular crop
plant. A brief review of the effects of self-fertilization with
several different crop plants will serve to show the wide diversity
of results when selfing is practiced and indicate the difficulty of a
close standardization of breeding methods. It seems probable
that such standardization will depend, in a large measure, on the
effects of self- and cross-fertilization with the crop plant in
question.
As has been emphasized in Chaps. Ill and XIV, it is probable
that more information is available on the effects of self-fertiliza-
tion in corn than for any other cross-pollinated plant. In
general, all inbred lines of corn obtained so far are less vigorous
than normal corn, although some inbred lines are relatively
vigorous and normal in habit of growth. Inbred lines differ
widely in resistance to diseases, such as bacterial wilt in sweet
corn and reaction to smut, as well as ability to withstand environ-
mental conditions generally considered unfavorable. The most
noticeable effect of inbreeding in corn, in addition to reduction of
vigor and the isolation of lines that are .relatively homozygous, is
the appearance of many recessive abnormalities.
244 METHODS OF PLANT BREEDING
Studies of inbreeding with alfalfa have been made by Kirk
(1927, 1932, 1933). In general, the loss in vigor is rather great
when alfalfa is self-fertilized for several generations. Kirk says,
*'The results of selfed line breeding have not been impressive as a
practical method of improvement." H. M. Tysdal, of the U.S.
Department of Agriculture, cooperating with the Nebraska
Agricultural Experiment Station, has continued self-fertilization
with alfalfa for a greater number of generations than has been
reported by any other investigator. The results are given here in
considerable detail and may be compared with those for corn that
have been discussed in Chap. III. From three trials under
Nebraska conditions using, in two of the three studies, lines of
alfalfa with known genetic characters, Tysdal concluded, "As an
average of the three tests under open-pollinated conditions, 89.1
percent natural crossing was found." This is somewhat lower
than in corn but higher than has usually been reported for
alfalfa.
The following statement from Tysdal (1941) describes the
results of a study of the effects of selfing on forage and seed yields :
From the amount of natural crossing found in alfalfa, it would be
expected that self-fertilization would lead to a reduction of vegetative
[,7owth and seed yield. The selfing program has been included for a
number of years as a part of the alfalfa improvement and breeding
program at Nebraska. While most of the lines have been selfed for
only one or two generations, a few have been carried into the seventh
and eighth generations of inbreeding. A number of these lines were
planted in space-planted nurseries with the rows spaced 27 inches apart
and the plants separated by 18 inches in the row. The self -fertilized
lines were planted in comparison with hybrids between these lines, their
open-pollinated progeny, and the standard varieties, Grimm, Ladak
and Hardistan, the latter belonging to the Turkestan group, which
represented the varieties from which practically all of the inbred lines
originated.
Yields of seed and of forage were obtained in terms of the
average yield of the three varieties. The forage yields were
obtained by taking green-weight yields on a 2-year basis and the
seed yields for a single season. Results given in Table 29 are an
average for the number of lines tested, all lines containing at least
10 plants. Us7m])y 30 V> 60 pla,nte formed the basis for taking
yields-
CONTROLLED POLLINATION METHODS
245
TABLE 29. YIELDS OF SELF-FERTILIZED LINES OF ALFALFA IN PERCENTAGE
OF THE PARENTAL OPEN-POLLINATED VARIETIES GRIMM, HARDISTAN,
AND LADAK*
Number
of selfed
Number
of lines
Actual yield in per cent
of original parents
Theoretical yield
generations
tested
Forage
Seed
Forage
Seed
1
54
68
62
68
62
2
17
48
39
52
43
3
9
59
38
44
33 5
4
13
51
36
40
28.75
5
1
41
29
38
26 37
6
37
25.18
7
1
26
15
36 5
24 58
8
4
28
* 8
36 25
24 28
of H. M. Tywdal.
IUU
90
to
o
ieo
6
I
fD
g. 6
IBQ
2-40
c
0>
8ZO
t-
0>
CL
EIO
T3
0>
I
\
x = Actual forage productivity
Upper curve - Theoretical tbrage productivity
- Actual seed productivity
Lower curve = Theoretical seed productivity
N
\
\\
)
(
(
\ v
\
x
^ -
J
^^^n
.
)
~-
X
<
4
I
6
23456
Number of selfed generations
FIG. 33. Point diagram showing average forage and seed yields of self-fertilized
lines of alfalfa after one to eight generations of inbreeding, and curves showing
theoretical decrease in yield of self-fertilized lines. (Courtesy of H. M. Tysdal.}
246 METHODS OF PLANT BREEDING
The theoretical yields given in the table were calculated by
comparing the average yields of the open-pollinated varieties
Grimm, Hardistan, and Ladak as 100 per cent with the yields of
54 first-year selfed lines at 68 per cent, giving a reduction of 32
per cent and the expectation that yields would be reduced one-
half in each succeeding generation. For example, subtracting
one-half of 32, or 16, from 68 gives an expected yield for second-
generation selfed lines of 52 per cent. These results are given in
the diagram in Fig. 33. Although selection seemed to lead to a
slowing up in the reduction in yields in the first five selfed genera-
tions, further years of selfing caused a greater reduction in yields
of both seed and forage than the theoretical expectation.
In discussing these results, Tysdal says, in part :
It would be difficult if not impossible to give an exact curve of reduc-
tion in yield caused by inbreeding in alfalfa because it would be necessary
to consider the origin of the material as well as to very carefully refrain
from any type of selection whatever. Obviously, selection is practiced
among the inbred lines in a breeding and improvement program and,
therefore, the results presented above are subject to whatever bias
may result from such selection. In some cases, lines were carried for
the purpose of determining the principle rather than for selective pur-
poses, but on the other hand, some were eliminated in the selection
program while still others reduced so rapidly in seed yield that they
could not be carried at all. To indicate the wide range in forage yield
of selfed lines, for example, it is only necessary to point out that the S\
lines varied from 26 per cent to 105 per cent. Seed yield is even more
variable in selfed lines than forage yield. Some lines in advanced
generations showed increased productivity over the original parent,
while others decreased very rapidly. This divergence might be attrib-
uted, at least to some extent, to the variability in seed setting in alfalfa
in general, and also perhaps, to the peculiarities of the conditions under
which the test is made. Those lines which might be selected for high
self-fertility, as autogamous lines, for example, might produce unusually
well under conditions of limited cross pollinating insect activity. The
origin of the selfed lines no doubt also plays an important part. Selfed
lines from Turkestan origin apparently do not reduce as rapidly as
those from Grimm or Ladak origin. In general the Turkestan group
appears to be more homozygous than alfalfas of hybrid origin such as
Ladak and Grimm, and it may be that the diversity of origin of the
latter would produce a greater range and possibly a different type of
curve in yields of inbred progenies. Further, when a given plant is
CONTROLLED POLLINATION METHODS 247
chosen for selfing from a mass population, there is no way of knowing
whether it, itself, was the result of cross- or self-fertilization.
The results obtained with alfalfa "are remarkably similar to
those obtained in corn and this together with the fact that hybrid
vigor has been demonstrated in alfalfa similar to that in corn
(unpublished data) leads to the conclusion that the principles of
breeding in this crop are essentially the same as those which have
been established for com."
With rye, Heribert-Nilsson (1916, 1919, 1921) found 1 or 2
plants out of 100 that were highly self-fertile, although self-steril-
ity was the usual condition. Some inbred strains approached in
yielding ability the normal variety from which they were
obtained. The number of recessive abnormalities is somewhat
less in rye than in corn. Brewbaker (1926) believed self-fertiliza-
tion and selection a desirable method of breeding rye, although
from studies that have been continued at Minnesota no inbred
lines have been obtained as vigorous as normal. Peterson (1934)
studied crosses between self -fertile and self-sterile inbred lines and
found that the factor for fertility bred out the sterility allelcs in
later generations of selfing from a cross between fertile and sterile
lines. Whether self-fertility is desirable or undesirable in rye is
an unanswered question.
With sunflowers, Hamilton (1926) found a reduction in vigor,
after selfing in most lines. He says, " Unlike the inbred strains of
corn, however, a number of the sunflower strains, while becoming
extremely uniform, did not lose any of their former vigor. In fact
some of the tallest, leafiest and highest yielding rows under test
during the past five years were strains that had been inbred for
five consecutive generations."
In timothy breeding, as developed at Cornell, selfing was
practiced for a year and the better types then used for breeding
stocks. Clarke (1927), at Minnesota, concluded that vigorous
lines of timothy could be obtained without great difficulty and
believed that self-fertilization and selection in self-fertilized lines
was a practical means of breeding. Valle (1931), in a breeding
study in Finland, states that the percentage of self-fertile and
self -vital lines was too low to make self-fertilization and selection
a valuable method of breeding timothy.
Jenkin (19316), at Aberystwyth, placed timothy in two main
groups, the hay type, with 42 chromosomes, and the pasture tyt)e,
248
METHODS OF PLANT BREEDING
with 14. Self-fertility was variable, the pasture type being much
less self-fertile than the hay type. Self-sterile plants are frequent
in commercial varieties of American timothy, although self-
fertility is common also. Two strains of timothy have been
FIG. 34. Individual timothy plants grown under like conditions. The upper
plants are undesirable, one having weak stems and the other lacking vigor. The
lower plants are more desirable. They differ in density of plant and number of
culms. (Courtesy of Myers.)
selected at the Minnesota station that were self -pollinated for a
3-year period. One of these strains is from a single self-pollinated
line and the other from a combination of several self-pollinated
lines. In preliminary comparisons with normal commercial
timothy, both strains appear somewhat superior to the commer-
cial variety. These and other results of a similar nature prove
that in some cases improved varieties can be obtained directly by
the utilization of self-pollinated lines from crops that presumably
belong to the cross-pollinated group. In brome grass, Bromus
CONTROLLED POLLINATION METHODS 249
inermtSj Kirk (1932) obtained a vigorous nonstooling line after
four generations of selfing.
Most perennial grasses show considerable self-sterility, and the
question of the desirability of selection in self -fertilized lines as a
method of breeding grasses is undecided at the present time.
Beddows (1931) has summarized many previous studies on seed
setting in the grasses and presented further data comparing seed
setting from enclosed inflorescences in relation to seed setting
under normal open-pollinated conditions in terms of a ratio of
heavy seeds per 100 spikelcts in free inflorescences (open-polli-
nated conditions) over enclosed (self -pollinated conditions).
Most annual species of grasses were rather highly self-fertile,
whereas many of the perennial species set much less seed on
enclosed inflorescences than under normal open pollination. In
perennial species of grasses, the ratio of seed setting of free over
enclosed, F/E, ranged from 0.95 in Agropyron tenerum to 152.95
in B. inermis. In Lolium perennc, Jenkin (193 la) studied several
different plants and clonal progeny extensively. Plant 43
produced only an average of 0.8 seeds per 100 spikelets when self-
pollinated, whereas plant 48 produced 117.6 seeds. Wholly self-
fertile lines were easy to obtain. With orchard grass, Dactylis
glomerata, Stapledon (1931) obtained some self-fertile and self-
vigorous plants, several inbred lines remaining vigorous after 5
years of selfing. On the average, however, the plants resulting
from selfing are much less vigorous than those derived from
crossing.
With red clover, Trifolium pratcnsc, Williams (1931a) found 3
plants out of 262 original plants to be truly self -fertile; he
explained self- and cross-incompatibility on the basis of an
extensive series of self -sterility alleles. White clover, T. repens,
was highly self-sterile according to Williams (19316), although
less so than red clover. Atwood (1940, 1941) has studied self-
and cross-incompatibility in T. repens and explains his results on
the basis of a multiple-allelic series of factors. One plant out of
615, when self -pollinated, set seed freely and may have carried a
factor for self-fertility,
A highly self-fertile line of red clover has been bred at the
Minnesota station, and in crosses between this line and normal
self-sterile plants fertility continues to be the dominant type
under self-pollination conditions. Groses between the self-
250 METHODS OF PLANT BREEDING
fertile line and normal plants were used as a basis for selection
under isolated normal open-pollinated conditions. The origin
consisted of 50 plants from commercial red clover crossed with the
self -fertile line. The plan of selection consisted of growing about
1000 plants each generation in a nursery individually spaced,
the selection of 100 vigorous, desirable plants, and the discarding
before flowering of other plants in the nursery, allowing cross-
pollination of the selected plants. Similar selection was made
in another isolated plot using normal commercial northern-grown
seed. After three generations of selection, the two types of origin
were compared by growing their progeny in rows, with the result
that the cross of the self -fertile with normal appeared less vigorous
than the selection from normal commercial red clover.
It is generally believed that the cucurbits, comprising cucum-
bers, muskmclons, watermelons, pumpkins, and squashes, belong
to the cross-pollinated group of plants. Rosa (1927) stated that
the amount of cross-pollination in melons varied with the variety
and ranged from 5 to 73 per cent. Whitaker and Jagger (1937)
state that cucumbers, squashes, and pumpkins are normally
strictly monoecious, whereas muskmelons and watermelons are
andromonoecious. Andromonoecious species bear bisexual or
complete flowers instead of strictly pistillate ones, in addition to
staminate flowers. When one considers the way that these
plants are grown and the necessity of insect pollination, it is
apparent that frequent cross-pollination must occur.
The cucurbits, as a group, show less reduction in vigor due to
inbreeding than most members of the cross-pollinated group of
plants. Bushnell (1922), Haber (1929), and Cummings and
Jenkins (1928) found no great loss in vigor as a result of ( con-
tinued self-pollination with squashes. Cummings and Jenkins
studied the effects of continued self-pollination for 10 generations.
Similar results were obtained by Porter (1933), Rosa (1927), and
Scott (1932) with watermelons. C. F. Poole reports (unpub-
lished) that lines of the Northern Sweet watermelon that have
been selfed for seven generations, show no reduction in size of
melon when compared with the commercial lines of the same
variety. Whitaker and Jagger concluded that hybrid vigor
probably did not occur in any of the cucurbits. There are, how-
ever, two recent reports by Hutchins (1938) with the cucumber
and by Curtis (1939) with summer squash that show marked
CONTROLLED POLLINATION METHODS 251
hybrid vigor. Hutchins suggested that it would be feasible to
utilize the hybrid vigor of F\ crosses in commercial production.
Curtis outlined a method for the production of hybrid seed
with summer squash, Cucurbita pepo, by growing the two
varieties to be crossed in alternate rows and the removal of all
male flowers from the seed variety before the male flowers have
opened.
INHERITANCE OF SELF-INCOMPATIBILITY
Crane and Lawrence (1934) draw a distinction between incom-
patibility and sterility. Incompatibility is due to some physio-
logical hindrance to fertilization. The pollen and ovules or at
least a good proportion of them are functional, the failure to
obtain seed being due to slow pollen-tube growth. Sterility is
classified by Crane and Lawrence into: " (1) generational sterility,
due to the failure of any of the processevS concerned with the
normal alternation of generations, namely, development of pollen,
embryo-sac, embryo and endosperm, and the relation of these to
one another and their parents regardless of the cross made and
(2) morphological sterility due to suppression or abortion of the
sex organs."
Many species of plants are often self-sterile, and among thes^
there are many plants of economic importance. These include
fruits, perennial grasses, rye, sorrie clovers, alfalfa, sugar beets,
some Brassica species, and some plants grown for ornamental
purposes. East (1929) and Brieger (1930) have given extensive
reviews of much of the literature.
Crane and Lawrence (1934) credit Prell (1921) with first
suggesting a genetic explanation of self-sterility on the basis of a,
series of self-sterility alleles and East and coworkers for the firsl
proof of such a series in Nicotiana. Self -incompatibility appear; ,
to be a somewhat more desirable term than self-sterility for those
cases where self-fertilization is prevented, although the pollen
grains and egg cells are functional.
There has been a rapid a* 'cumulation of information regarding
self -incompatibility in recent years, and in many species rather
clear demarcation between self-compatibility and self-incom-
patibility. In other cases, the differences are not so clear-
cut, and there may be a gradual graduation from self -fertility to
self-sterility through a series of causes. Crane and Lawrence
252 METHODS OF PLANT BREEDING
have given evidence for the conclusion that in some cases this
may be the result of polyploidy and the duplication of several
series of sterility alleles.
Two types of inheritance seem of general interest. The oppo-
sitional-factor hypothesis furnished a satisfactory explanation of
self- and cross-incompatibility in tobacco by East and coworkers.
The genes responsible belong to a series designated by S, and like
other alleles, two factors may be carried by a single diploid plant,
a series of 15 such allelic factors having been found in tobacco.
A pollen tube carrying any one of these alleles, Si to $15,
shows slow pollen-tube growth in the stylar tissue carrying the
same factor but normal pollen-tube growth in stylar tissue carry-
ing a different genetic factor for self-incompatibility. A factor
for self-fertility S/ was found, also, that was functional with any
of the 81 to $15 alleles, and self-fertility was dominant to sterility
"in crosses. Self-fertility of this nature would breed out incom-
patibility in the selfed progeny of crosses, which would make it
possible to add the factor for fertility if desired.
Types of results that may be expected will be illustrated
briefly. Parental genotypes FI, F% and backcross progeny in a
diploid organism make clear the types of breeding behavior.
Two self-sterile plants are used as parents in the hypothetical
illustrations, their genotypes being 8183 and 8^8*. When SiS$
is self-pollinated, seed production does not commonly result,
since pollen tubes carrying either of the alleles Si or 83 grow too
slowly in stylar tissue of the same genotype. Exceptions to this
rule have occurred in many species of self-sterile plants leading
to homozygous individuals of the genotype SiSi and 8383, but
seed is too infrequent to make this method of seed production
efficient as a means of controlled selection in self-pollinated lines
of self-sterile species. With some species such as cabbage,
Brassica oleraceae, controlled bud pollination of a self-sterile
plant gives good seed production, and selection in self-sterile
lines may be practiced, leading to the isolation of homozygous
lines. A cross of 8183 X $2$4 produces four types of FI offspring
SiS%j SiS^j 8283, Ss$4. Each of the types is self-sterile but fertile
with their parents in backcrosses and with each other. A cross
of 8182, X SiSi, for example, will produce plants that are SiSt
and 8284. The reciprocal cross of 8184 X 8183 will produce
8183 and S s St.
CONTROLLED POLLINATION METHODS
253
When the self-fertility allele S/ is present, continued self-
pollination leads to the rapid isolation of self-fertile genotypes.
For example, S/S f X SiS* -* S f Si and S/S 8 . If S f Si is self-
pollinated, only two genotypes result, one like the parent SfSi,
$183x8254
A B C D
FIG. t35. Diagrammatic representation of pollen-tube growth in compatible
and incompatible crosses, (a) and (b) Incompatible, slow pollen-tube growth;
(c) compatible, all pollen able to effect fertilization; (d) only $2 pollen functional.
the other homozygous for S/S/. The pollen tube developing
from a pollen grain carrying Si makes slow growth in stylar tissue
OS/Si) carrying the self -sterility allele Si.
Riley (1934, 1936) has explained self-sterility in Capsella
grandiflora, a diploid species with eight haploid chromosomes, on
the basis of the sporophytic nature of the parent plants and on the
interaction of two pairs of genes. Before giving the genetic
explanation, a brief summary of the results of the crosses will be
given.
Three intrasterile, interfertile classes have been found in
C. grandiflora. These have been designated classes A, C, and B.
Class A X class C will produce classes A and C, or A, jB, and C,
but never A and B only. Class A X class B will produce classes
A and B, or classes A, B, and C or classes A and C. Class
B X class C will produce classes B and (7, or class C only, but
never class A. Reciprocal crosses give the same result.
When a plant of C. grandiflora was crossed with any one of the
three self-fertile species of Capsella, the FI was fully fertile and
completely self-compatible. Segregation for self-fertility and
self-incompatibility occurred in the F%.
254
METHODS OF PLANT BREEDING
The following table (Riley 1936) gives the genetic explanation
of results from crosses within and between the three self-incom-
patible groups.
TABLE 30. STERILITY AND FERTILITY IN CROSSES WITHIN AND BETWEEN
THE THREE SELF-INCOMPATIBLE GROUPS OF C. grandiflora
Cltu
38 A
Class C
Class B
Genotype
TtS c S c
Tt&s
Ttss
US C S C
ttS*s
ttss
Tt$ c S c
S
S
S
F
F
F
TtS c s
s
S
S
F
F
F
Ttss
8
S
s
F
F
F
ttS e S
F
F
F
S
S
F
ttS c s
F
F
F
S
S
F
ttss
F
F
F
F
F
S
S sterile, F = feitile.
These results were explained on the basis of the sporophytic
nature of the parent plants. Members of class A were incom-
patible in crosses together, because they possess the dominant
gene T, which is epistatic to S c . Any plant that is homozygous
for the recessive condition, or tt, is fertile in reciprocal crosses with
members of class A. Factor T is never in the homozygous
condition, since plants bearing it can be crossed only with the
homozygous recessive tt.
A second pair of genes S c s is responsible for the differentiation
of classes C and B. Class C carries this pair of genes either in the
homozygous condition S C S C or the heterozygous condition S c s,
and class J5 is homozygous for the recessive condition ss.
A self-compatible factor S* is a member of the $ series and
dominant to S c or s and epistatic to T 9 and S c and s are hypostatic
to T.
East (1934) suggested that substances in the stylar tissue of
Nicotiana react with substances in the pollen tube to slow up the
rate of the pollen tube growth in self -incompatible combinations.
The further suggestion was made that, in most plants, these sub-
stances are not present in the young bud but appear during the
24 hr. preceding the opening of the flower. In self-sterile geno-
types that are self-fertile when pollinated 24 to 48 hr. before the
flowers open it was presumed that these substances were not
CONTROLLED POLLINATION METHODS 255
produced during the 24 hr. preceding flower opening. In another
type in which self-sterile plants are self-fertile at the end of the
flowering period it was presumed that these plants were unable
to produce an adequate amount of the inhibiting substance late
in the flowering period. The inhibitory effect appears to be
localized in a certain region of the style, since the growth rate
of the pollen tube is slowed up markedly as the pollen tubes
reach this place, but after this region is passed the growth rate
again approaches normal. In Capsella, the pollen grains in self-
incompatible matings do not germinate, or produce tiny abortive
tubes only. This has led Riley to suggest that these inhibitory
substances are located at the very end of the stigma in the
stigmatic hairs. Yasuda (1934) believes these inhibitory
substances originate in the ovule in petunia, from which they
may ascend the stigma, depending on genetic differences in the
plants and the environmental conditions under which they are
grown. If they reach the stigma, they may inhibit germination
of the pollen grains. If they only reach the style, they may
inhibit pollen-tube growth in the stylar region, and in some cases
these substances stay in the ovary and inhibit pollen-tube growth
in the ovary. The suggestion is made also that weak self-
fertility may be the result of a low production of these inhibitory
substances.
Studies in Wisconsin by Brink and Cooper (1939) and Cooper
and Brink (1940) with alfalfa deal with partial self-incom-
patibility, the type of self-sterility that probably occurs very
commonly in many crop plants. Cross-pollination produced a
much higher average number of seeds per flower than self-
pollination. In a comparison of seven plants selfed, with crosses
between them, 14.6 per cent of self-pollinated ovules were
fertile, whereas 66.2 per cent of cross-pollinations led to fertiliza-
tion. A low degree of fertilization under conditions of self-
pollination was believed to be explained primarily on the basis
of the oppositional-factor hypothesis.
Of the ovules that became fertile, 34.4 per cent containing
inbred embryos and endosperms collapsed within 6 days after
fertilization; in the cross-pollinated plants, only 7.1 per cent,
containing hybrid endosperms and embryos, collapsed. These
differences are highly significant and seem to be dependent upon
the relative rate of growth of the endosperm tissue and embryo.
256 METHODS OF PLANT BREEDING
The collapse of ovules during the early-development stages after
fertilization has been called somatoplastic sterility by Brink and
Cooper. They say:
The embryo sac in the mature ovule of alfalfa is surrounded by two
integuments. The inner integument, which is composed of two layers
of cells, lies in direct contact with the embryo sac except at the chalazal
end where a few disintegrating cells, remnants of the nucellus, are found.
Shortly after fertilization active cell division is initiated in the integu-
ments as well as in the endosperm mother cell and the zygote.
The critical factor for survival seems to be the manner in which the
translocated food is shared between the endosperm, on the one hand,
and the inner integument, on the other. The partition of nutrients
appears to depend upon the rate of growth inside and outside the
embryo sac.
The endosperm is considered to be the dominant tissue within
the embryo sac. When the endosperm keeps pace in its growth
with the surrounding material, tissue development of the seed
continues in a normal mariner. The rate of growth of the embryo
is much slower than that of the endosperm and not very different
in hybrid than in inbred embryos. The writers say, "The initial
conditions iri the ovule outside the embryo sac being alike in the
two cases, it seems clear that the higher survival following
crossing is the result of the more active growth of the hybrid
endosperm. Conversely, following self-fertilization, the rate of
growth of the endosperm is frequently so low that the balance
soon shifts in favor of the integuments."
Other cases were noted by Brink and Cooper from species
crosses, where fertilization had taken place but early collapse after
fertilization followed because of slow growth of the endosperm
and the lack of nutrients for the growing embryo. Although
these results have not been placed directly on a genetic basis,
it seems probable that they result from causes similar to those
responsible for the reduction of vigor in selfed lines of cross-
pollinated plants. If one accepts the Mendelian explanation of
heterosis, then it seerns probable that the early collapse of ovules
after fertilization is due primarily to genetic causes.
Other cases of self-sterility are known. Heterostylism, i.e.,
differences in relative length of the styles and stamens, may cause
a lack of seed production under conditions of self-pollination.
CONTROLLED POLLINATION METHODS 257
Proterandry or proterogyny also may be causes of cross-pollina-
tion and make self-fertilization difficult.
Although much is known regarding problems of self-sterility,
different investigators have reached widely different conclusions
regarding the possibilities of using controlled self-pollination
as an aid in breeding. These differences in opinion are doubtless
dependent upon species and varietal differences in response to
self-pollination, or the effects of differences in environmental
conditions. The extent to which controlled self-pollination can
be used in breeding remains an unanswered question in many
cases.
METHODS OF BREEDING
Certain principles rather generally accepted as of importance
in relation to the controlled method of breeding in corn appear
to be applicable to other cross-pollinated crops. Certain of
these may be restated.
1. Yield and many other characters of economic importance are
the end result of the interaction of multiple factors.
2. Inbred lines show wide genetic differences and, as a rule, are
less vigorous than the normally pollinated varieties from which
they originated.
3. Inbred lines differ in combining ability in crosses. Although
there is considerable evidence that combining ability in crosses is
positively and significantly correlated with those characters that
are expressions of vigor in the inbred, it is equally evident that,
of two inbred lines that in themselves seem equally desirable, one
may give much greater vigor than the other, on the average, in
crosses with unrelated inbreds.
4. Combining ability of an inbred may be tested by crossing
it with a commercial variety, i.e., the top-cross is a relatively
satisfactory method of learning relative combining ability.
Crosses with a series of inbreds used as testers is another method
now in use by the corn breeder for testing combining ability.
Some investigators have advocated the use of testers that in
themselves are undesirable for important characters.
5. Combining ability is inherited in much the same manner as
other quantitative characters. If two low combining lines are
crossed and selection practiced during the segregating generation
under controlled self-pollination until relative homozygosis is
258 METHODS OF PLANT BREEDING
obtained, most of the resulting inbreds will be low in combining
ability. Conversely, inbreds selected from a cross of high-
combining inbreds are mostly high in combining ability.
6. Genetic diversity is of importance in relation to heterosis.
Crosses between inbreds from a different origin show greater
heterosis, on the average, than from a related origin.
7. Combining ability may be determined during the early
generations of selfing by means of the crossing test.
When controlled cross-pollination can be carried out on an
extensive scale at a reasonable cost, it seems that the method
of breeding by controlled pollination that has been developed
for corn can be applied directly to other plants of the cross-
pollinated group. In some cases, it seems feasible to introduce
a factor for male sterility in one of two inbred lines that are to
be used to produce F\ crossed seed for commercial seed produc-
tion. The two lines to be crossed would be intcrplanted, and all
male fertile plants would be removed before pollination. With
perennial plants, the same field could be used for several years.
Further increases for seed production could be made by vegeta-
tive propagation.
A method suggested by Pearson (1932) for breeding cabbage is
of interest where self-incompatibility is of common occurrence
and controlled bud pollination leads to the production of a con-
siderable amount of selfed seed. In cabbages, both self-fertile
and self-incompatible lines may be obtained. Pearson suggests
the selection and isolation of self-incompatible lines by pollinating
in the bud stage. These self-incompatible lines can be differ-
entiated from the self-fertile by pollinating at anthesis also and
discarding the lines that are self-fertile, i.e., lines that set seed
when pollinated at anthesis. After a considerable number of
self-incompatible lines have been selected, these then may be
tested for combining ability. Although not suggested by
Pearson, it seems that the inbred- variety cross method would
be desirable to use in the first elimination of lines, with the use
in the crossing study of only those lines that proved to be good
combiners. After selecting the best inbred lines by this means,
artificial crosses between lines should be made, making the
crosses at anthesis. Those crosses that set seed would then
be tested for producing ability. After obtaining a desirable
cross, it could be maintained by bud pollination of the parent
CONTROLLED POLLINATION METHODS 259
lines and the continued production of controlled cross-pollin-
ated seed by interplanting members of the two lines for seed
production.
A plan of breeding grasses originally adopted at Aberystwyth,
Wales, and in New Zealand (Levy 1933) is of general interest.
It consists of collecting material from its natural habitat, from
foreign sources, and from other breeders, and of making a study
of several thousand individual plants, perhaps with the final
selection of not more than 100 or 200 out of 5000. The progeny
of these selected plants may be increased and given further study,
with the use of one of several methods of breeding, depending on
the nature of the material, the possibility of self-fertilization,
and the extent to which it is possible to isolate vigorous self-
fertile lines. In many cases, a rapid increase of material from
the selected plants may be desirable. A simple method consists
of interplanting the clonal progeny of these selected plants,
allowing them to cross by natural means. When facilities are
available for an intensive breeding program with a particular
species, desirable-appearing clonal lines or closely bred lines that
may or may not have been previously bred by controlled self-
pollinatiori may be used as a basis for breeding of improved
varieties. By means of brother-sister mating or diallel crossing
and the test of crosses in F\ and in F 2 , the more desirable progenies
may be isolated and combined to produce improved synthetic
varieties.
Some investigators are using the so-called Macauley (1928)
method as a means of isolating relatively homozygous lines, where
controlled self-pollination, because of self -sterility, or where
reduction in yield as a result of continued selfing is so great that
the isolation of selfed lines does not seem desirable. Macauley
suggested a method of close breeding for corn that may be
adapted for this purpose. As applied to corn, it consists of grow-
ing the progeny of selected ears each in an individual plant plot
of approximately 200 plants, preventing cross-pollination between
plots by means of natural barriers such as the use of border rows
of a much later maturing corn and thus forcing pollination within
each plot. The more desirable plots are selected each generation
and several ears again selected to plant the isolated plots for
the next generation. It was concluded on theoretical bases that
four or five generations of this sort of selection would be equiva-
260 METHODS OF PLANT BREEDING
lent, in an approach to homozygosity, to a single generation of
selfing. The inbred lines obtained by this method of breeding,
where sufficient vigor is retained, could be used directly as an
improved variety, or several lines could be combined to produce a
synthetic variety.
When intensive studies of improvement have not been made
previously, the recognition arid propagation of improved ecotypes
that have developed through natural selection may prove of
value. By these methods, a considerable scries of new strains
of great value have been developed in New Zealand. These
include Hawks Bay and Poverty Bay perennial rye grass, Akaroa
cocksfoot (orchard grass), New Zealand white and New Zealand
extra-late red clover, Marlborough lucern, and New Zealand
brown top.
In spite of some of the difficulties, it seems advantageous to
outline methods of breeding. These are based to a considerable
extent on methods found applicable to corn.
OUTLINE FOR IMPROVEMENT OF CROSS-POLLINATED PLANTS BY CONTROLLED
POLLINATION METHODS
I. Selection in self-pollinated lines.
A. In general, use adapted varieties, arid artificially self-pollinate as
many plants as can be handled with the available facilities;
unadapted varieties may be used if any desirable character is
wanted.
B. Grow the progeny of each self-pollinated plant from self-fertilized
seed. The number of plants in each selfed line of the first and suc-
ceeding generations should be sufficient to give an adequate sample
of the progeny. This number ordinarily should not be less than 20,
arid a minimum of 30 to 40 plants is desirable. In many crops it
will be advantageous to start the seedlings in the greenhouse and
transplant into the fields. The plants should be spaced far enough
apart to permit individual study.
C. Self-pollinate one or more desirable-appearing plants from each
desirable first-generation selfed line. In general, plants of at least
average vigor should be selected for selfing.
D. Following the procedure in (7, grow successive generations of selfed
lines until relative uniformity is secured. As the process of selec-
tion proceeds from first to later generations, greater weight should
be given to vigor of growth and more emphasis placed on high
fertility. As elimination of weak and undesirable lines is effected,
their place in the nursery may be filled with additional selections
from the strong desirable lines, with new first-generation selfed lines,
or by selections from crosses between selfed lines.
CONTROLLED POLLINATION METHODS 261
E. Any selfed lines of promise may be tested for reaction to disease in
a special disease garden, or a disease epiphytotic may be induced
in the self-fertilization plots.
II. Improvement of selfed Unas.
A. Backcrossing. A desirable method when one wishes to retain all
or nearly all the characteristics of one line and add some char-
acters to it. Easy of accomplishment when the character to be
added is inherited in a relatively simple manner.
Examples from corn:
1. To add yellow endosperm to a selfed line that is desirable
in other characteristics and is breeding true for white endo-
sperm.
2. To add tender pericarp to a sweet-corn variety or selfed line
that has tough pericarp.
B. Convergent improvement. A desirable method of increasing the
vigor of each of two desirable selfed lines that combine well in an FI
hybrid without modifying their combining ability.
C. The pedigree method. Select selfed lines as parents that have
complementary characters, i.e., lines that excel in different desirable
characters. After making the cross, selection during several
generations of selfing is practiced until practical homozygosity is
reached. Example: One parent with strong stalk, i.e., ability to
withstand lodging, the other with good general vigor but weak
stalk. Make the cross, and self-pollinate, and select during the
segregating generations.
III. Use of selfed lines as breeding material.
A. 1. Top-crosses or inbred sire crosses, an inbred line crossed with a
variety.
a. Of value in some cases as a commercial hybrid, or as a basis
for the selection of an improved clon in such crops as potatoes.
6. A desirable method of testing the combining ability of selfed
lines. By this means, the more promising lines are selected
to test in single, three-way, double crosses or in the production
of a synthetic variety.
2. Selfed lines, if sufficiently vigorous, may be increased for use as a
new variety.
3. Single crosses between two selfed lines may be made and the
cross grown as the commercial crop, providing the selfed lines
are sufficiently good seed producers.
4. Double crosses between two single-crossed hybrids may be made
and the cross grown as the commercial crop. Advanced-
generation single crosses may be used, particularly as the male
parent.
5. Three-way crosses may be used as the commercial crop. An FI
cross of two selfed lines may be used as the female and a selfed
line as the male parent.
6. New varieties may be synthesized by composite crossing of
several inbred lines or in special cases from two lines.
262 METHODS OF PLANT BREEDING
B. Compare new varieties and Fi hybrids with standard varieties by
means of replicated field-plot trials for a sufficient length of time to
establish their value.
C. Production of seed of F\. hybrids and new varieties by increasing
seed of improved varieties and inbred lines in isolated plots.
IV. Crop plants in which self-sterility is a factor.
A. Selection of self-fertile lines and the addition of the factor for
fertility to many lines of the crop plant, i.e., the breeding of self-
fertile lines and their use later as Fi hybrids or their combination
into synthetic varieties.
B. Self-pollination for a generation or two until some desirable char-
acter is hornozygous, followed by the combination of several lines.
(7. Selection in self-sterile lines, pseudofertile, by pollination in the
bud stage or other means such as the self-pollination of many
flowers and the use of the few seeds obtained as a means of selection
in normally self-sterile lines.
I). Cross-mating of plants selected on the basis of outstanding char-
acteristics. Progeny of crosses selected and better types isolated.
1. Strain building on the basis of selection of several plants as
parents and their bulk crossing by natural means. A broad
system of mass selection is practiced. This method is
applicable to the rapid increase of ecotypes that have devel-
oped by natural selection under a particular set of environ-
mental conditions.
2. Strain building by brother-sister mating.
3. Diallel crossing by hand. New strain developed from several
of the better crosses.
In diallel crossing, the breeder will have available inbred,
closely bred, or clonal lines that have outstanding characters.
After the crosses have been made, it may be desirable to test
their progeny in F\ and F z . In some cases grazing trials or other
tests of F% crosses will be helpful. Crosses will be combined in a
synthetic variety from parent plants or lines that combine well
with other lines to be used in the synthetic variety. It will be
possible, in many cases, to use Fi crosses to make the first combi-
nation of selected lines for use in the improved variety.
CHAPTER XVII
SEED PRODUCTION
The breeding of improved varieties of crop plants is carried on,
as a rule, by specialists who are trained in plant-breeding methods
and who have a knowledge of the needs of the grower and con-
sumer. Although a seed producer may undertake the problem
of breeding in some cases, the primary task of the seedsman
will be to produce high-quality seed of varieties and strains of
known value.
Good seed of any farm crop must be produced from a variety
or strain that is superior, insofar as that is possible, in the
following respects:
1. Adaptability to the locality and soil.
2. Purity of type.
3. Yielding ability.
4. Desirable agronomic characters.
5. Disease and insect resistance.
6. Quality for particular characters.
The seed of this adapted variety must be superior in the follow-
ing characters:
1. Germinating ability.
2. Color of seed and seed weight.
3. Uniformity.
4. Freedom from seed-borne diseases.
5. Freedom from noxious and other weeds.
6. Freedom from other damage.
7. Freedom from mixtures with other varieties.
These characteristics of good seed are, in general, appreciated
by seed growers. The first step in the production of good seed
is the selection of the variety or varieties to be grown.
SELECTING THE VARIETY
Improved varieties of farm crops bred by investigators at
federal or state agricultural experiment stations, including
varieties of wheat, oats, barley, and cotton, are registered through
263
264 METHODS OF PLANT BREEDING
a cooperative agreement by the Bureau of Plant Industry of the
U.S. Department of Agriculture and the American Society of
Agronomy. Registration is under the direction of a committee
of the agronomy society and is based on information from yield
trials, carried on for at least 3 years, in comparison with standard
varieties at federal or state agricultural experiment stations.
To be eligible for registration, a variety must be significantly
superior to the standard in some important character or charac-
ters and equal in other important characters. Registration <
consists of giving the new variety a register number and publish-
ing a description of its origin and characteristics in the Journal
of the American Society of Agronomy. Plant and seed samples
are furnished by the person or institution submitting the request
for registration.
Many of the state agricultural experiment stations list recom-
mended varieties and describe the conditions under which the
varieties usually give the most satisfactory performance. These
lists are based on actual field trials conducted on experiment-
station or farmers 7 fields in comparison with standard varieties.
The University of Minnesota Agricultural Extension Folder 22,
1941, revised whenever it seems necessary, gives the general
principles used in Minnesota in drawing up the recommendations.
Somewhat similar methods are used in other states. A variety
is added or removed from the recommended list of the Minnesota
Agricultural Experiment Station by vote at an agronomy con-
ference held each year. The following statement is quoted from
the folder.
The list of recommended varieties for Minnesota has the joint
approval of agronomists, plant breeders, arid plant pathologists of the
central experiment station at St. Paul and of the superintendents and
agronomists of the various branch stations at Waseca, Morris, Crooks-
ton, Grand Rapids, and Duluth. A variety must have been tested in
experimental plots for at least three years to be eligible for recommenda-
tion. The basis of recommendation is satisfactory performance in
competitive trials when compared with standard varieties. These
tests are conducted at the central and branch stations, in cooperative
trials on farms, and, in addition, comparative trials of reaction to
disease are conducted in specially prepared disease nurseries at the
central station. Varieties introduced from outside the state are given
the same careful trial as those developed in Minnesota.
SEED PRODUCTION 205
The list is followed by a statement of the important characters of
each recommended variety and its origin and regional adaptation. A
brief statement of varieties that are riot recommended is also given.
In Canada, the Canadian Seed Growers' Association has
accepted responsibility for deciding which varieties shall be
eligible for the production of certified or registered seed. In
general, varieties are accepted on the basis of their performance
in adequately conducted field trials in comparison with standard
varieties. The list of varieties eligible for use in the production
of registered seed is an important means in Canada of selecting
varieties for particular conditions.
Many of the states in the United States have crop-improve-
ment societies composed of growers interested in problems of seed
production. In some cases, the state association may select
varieties eligible for seed certification. These varieties are
chiefly those recommended by the state experiment stations,
although in some cases a few varieties in addition are selected by
the varietal committee of the crop-improvement society.
Large and small seed companies may in some cases breed or
select an improved variety. Improved varieties are described
in seed catalogues, which helps to make the characters of varieties
known to the general public. Many of the corn hybrids used in
the corn belt are produced and introduced by seed companies.
These companies use inbred lines of their own breeding together;
with those released by federal or state workers and introduce and
sell seed of hybrids for commercial growing under their own
pedigree, which in many cases is kept secret, although the
pedigree must be filed with a state official to comply with certain
state laws. In other states, all that is required, in addition to
the usual information required by seed laws, is a statement of
the type of hybrid and average days required to mature the
hybrid in various sections of the state.
Extensive yield trials of commercial seed-company hybrids,
in comparison with federal and state-experiment-station hybrids,
are made annually by most of the corn-belt experiment stations.
These trials are under the supervision of the Agricultural Exten-
sion Division and the State Agricultural Experiment Station*
An entrance fee is charged for each commercial hybrid grown in
the trials. Reports of the results of these trials are used by
266 METHODS OF PLANT BREEDING
growers as a means of selecting the hybrid that is best adapted
to their conditions.
The value of new varieties and their characteristics are brought
to the attention of growers by holding field days at the various
experiment-station fields at or just prior to harvest time, when
yield trials are discussed and the characters of particular varieties
may be observed by the grower.
In spite of the various methods used to inform the grower of the
relative merits of different varieties a large amount of seed of
overexploited and unadapted varieties is sold annually by
seedsmen. The loss could be done away with by a greater effort
to inform the producer of varietal characteristics and by a wider
use by the farmer of the information now available in the hands
of the state agricultural colleges, the experiment stations, and the
agricultural extension service,
FIRST INCREASE OF SEED OF A NEW VARIETY
A large proportion of new varieties of farm crops in the United
States are bred by state or federal investigators at agricultural
experiment stations. After deciding to recommend a new
variety, the problem of increase of seed and introduction of the
variety becomes of major importance. Many of the state experi-
ment stations keep on hand a small amount of pure seed of all
recommended varieties that serves as an initial source of pure
seed supply. First increases of new varieties of crop plants
often are made on experiment-station fields. Subsequent
increase is usually in the hands of seed growers who are members
of the state crop-improvement association.
After making the initial increase of seed of a new variety at
the state experiment station, further increase is made in Minne-
sota by the so-called " approved grower plan." A committee
of agricultural experiment-station workers decides how much
of the available seed supply will be distributed in each county.
Growers are selected by a county committee consisting of the
extension agronomist, county agent, and three farmers appointed
by the president of the Minnesota Crop Improvement Associa-
tion. The "approved grower" of registered seed is one who has
the following qualifications:
1. Willingness to cooperate to the fullest extent with the experi-
ment station, extension service, and other agencies interested in
pure seed production.
SEED PRODUCTION 267
2. Available clean land for seed production.
3. Facilities for storing seed so that mixtures may be avoided.
4. Previous satisfactory record in crop-improvement work
and in the community of which he is a member.
These approved growers purchase seed from the experiment
station and agree to place at the disposal of the experiment
station, if requested, at a price agreed upon by the seed-distribu-
tion committee, all seed over arid above that agreed upon for
use on his own farm. After this initial increase in the hands of
approved growers, subsequent seed increase is made by other
seed producers of the Minnesota Crop Improvement Association
or by others interested in seed production.
It was emphasized in the chapter on Corn Breeding that three-
way and double crosses in field corn were used chiefly by the
commercial corn grower. The propagation of inbred lines,
single crosses, and the production of three-way or double crosses
by the seed grower are essential phases of seed-corn production.
The larger seed companies take care of the initial increase and
maintenance of purity of the inbred lines used in their own
pedigrees.
There are two rather distinct plans that are followed for
hybrids released by state or federal workers. One method
consists of sales of small quantities of pure seed of inbred lines to
seed producers who make the subsequent increase of the inbreds
and single crosses necessary for three-way or double-cross seed
production. Some private breeders specialize in producing
seed of single crosses. All state experiment stations in the corn
belt have adopted a plan of initial seed increase of inbreds and
single crosses for newly released hybrids. Most of the experi-
ment stations release the inbreds after 2 or 3 years have elapsed
since the new hybrid was first released. Minnesota and Wis-
consin have developed methods for the increase of inbreds and
first crosses in sufficient quantity for the needs of all corn-
seed producers in their respective states. Similar work is
carried out in Ohio by a cooperative organization of seed
growers. Inbreds are not released for the hybrids that have been
bred and recommended by station workers in Wisconsin and
Minnesota.
The Minnesota studies have led to the conclusion that con-
siderable care must be taken to maintain the purity of inbred
lines. The methods used and some of the conclusions reached
268 METHODS OF PLANT BREEDING
have been summarized by Borgeson and Hayes (1941). The plan
now used is outlined as given by these writers.
Hand-crossed and selfed seed of all inbred lines needed in the corn
"program is planted each year in foundation plots at both the Southeast
and Central stations. JThe crop risk is distributed as much as possible
by planting at two stations with several dates of planting at each loca-
tion. Sufficient sclfed ears are produced to provide the necessary seed
that is needed the following year in the crossing plots where single
crosses are produced. The soiled ears are inspected both before and
after drying.
The seed is harvested and dried in fine-meshed bags in tray driers.
Twenty to 30 individual representative selfed ears of each culture are
saved and the balance of the sol fed seed bulked to use for producing
single crosses. Short "ear-to-row" cultures from 20 to 30 selfed ears
of each inbred are planted also in the foundation plot. Hand crosses
are made between the individual ear cultures obtaining several crossed
ears from each combination of "ear-to-row" cultures as follows: 1X2,
2X3, 3X4, etc., where 1 to 4, etc., represent the "ear-to-row"
cultures of each of the inbred lines, respectively. The hand-crossed
ears in each culture are examined and desirable crosses are bulked using
representative cultures. The crossed bulked seed is used the following
year as the parental source for the rather extensive hand sclfing program
that furnishes the major source of selfed seed for single cross increases.
In some seasons it is necessary to use hand controlled sib-pollination
when self pollination for any reason in some lines does not prove feasible.
The plan then consists of alternately producing selfed and crossed seed
for each inbred line.
The major features of the plan may be summarized briefly as follows:
When an inbred line seems relatively homozygous, sufficient selfed seed
of each inbred is produced each year to plant the necessary single cross
plots the following year. The seed planted for the selling plot is
obtained the preceding year from hand-pollinated crosses made by
crossing the progeny of "ear-to-row" cultures within the inbred lines
produced from selfed ears.
The rapid increase in demand for hybrid parent stocks made it neces-
sary to produce the larger part of the single crosses with individual
farmers. Two types of contracts have been used, one calling for an
acre rental fee and the other on the basis of production usually by
pound units. On the acre basis it did not appear that growers were
sufficiently interested in all cases to place the work on a desirable basis.
There was no great incentive to produce a high yield. On the other
hand, the yields of the single crosses were unpredictable, and it waa
difficult to arrive at a satisfactory price on the pound basis,
SEED PRODUCTION 269
This year a new form of contract has been used with the majority of
the growers. The grower is permitted to retain a share of the seed
stocks produced for his part in the contract. The balance of the seed
is then turned over to the experiment station for sale to other growers.
Small plots used for the increase of advanced generation seed are con-
tracted on the acre rental or unit payment plan.
During the past season it was possible to examine the first results of
the new methods of seed increase. All the single crossing plots were
planted with hand-pollinated seed. The purity of the parent lines
was highly satisfactory. Actual counts showed the percentage of off-
type plants to run on an average of about 0.25%, or 1 to 400. Any
rogues present were removed prior to tasseling.
From previous experience it is believed that it will be necessary to
provide hand-pollinated seed every year for the single cross plots. From
indications to date, the methods of increase outlined in this article
should provide both the purity and quantity of inbred seed desired.
SEED CERTIFICATION AND REGISTRATION
The Canadian Seed Growers' Association. The Canadian
Seed Growers' Association, first organized early in the present
century, was modeled after the Swedish Seed Association that
was first established in 1886. The objects of the Canadian associ-
ation, which was incorporated in 1920, may be made clear by
quoting from its Letters Patent, as given by Wiener (1937).
(a) Advancing the interest of Canadian agriculture by encouraging
seed growers and farmer members to maintain a high standard of
excellence.
(b) Developing the standards of quality for varieties and strains that
shall be eligible for registration.
(c) Establishing and maintaining a record of these varieties and
strains that are approved for registration.
(d) Fixing standards for the different classes of propagating stock of
varieties and strains that may be eligible for registration.
(e) Making provision for the necessary inspection of field crops and
propagating stock.
(/) Maintaining records of registered propagating stocks produced by
members.
({/) Encouraging the development and introduction of superior
varieties and strains.
(h) Providing for the multiplication and dissemination of propagating
stock of new varieties approved for registration,
270 METHODS OF PLANT BREEDING
(i) Co-ordinating the endeavors of plant breeders and seed grower
members of the association with the endeavors of crop producers in
general.
(f) Utilizing propaganda, advertisement and any other legitimate
means to increase the use of registered propagating stock.
(k) Developing a home market and if necessary an export market
for the disposal of surplus stocks.
(I) Such other means as may be found expedient from time to time.
The functions include the selection of varieties that are eligible
for the production of registered seed, the production of foundation
or elite stock seed of these varieties, the production by individual
members of registered seed for sale to commercial growers, such
seed having been sealed as registered seed after field and bin
inspection.
The work of the association is made possible by an appropria-
tion of the Dominion government, donations from companies
interested in high-quality productions, and through the help of
Dominion and Provincial organizations interested and engaged
in various phases of the crop-improvment projects.
The methods developed in Canada have been used rather
extensively as a basis for seed certification in various states in the
United States. This work has been carried out through the vari-
ous state organizations. Uniformity in methods has been devel-
oped through the International Crop Improvement Association.
The International Crop Improvement Association. This
association was organized in 1919 at a meeting in Chicago of
representatives of the Canadian Seed Growers' Association and
of state crop-improvement associations in the United States.
The object of the association can be adequately understood by
quoting from the constitution, where the purpose is stated to be
... to promote the agricultural interests of the various states and
provinces of America, emphasizing especially the improvement of field
crops in general and seed improvement in particular by:
(a) Encouraging the breeding and improvement of field crops and
seeds.
(h) Husbanding, propagating, and disseminating Elite, Registered,
Certified and Improved seeds.
(c) Creating a more active interest in better seeds through circulars,
reports, and otl er publicity, as Avell as encouraging local, state and
international shows.
(d) Assisting in the standardization of the seed improvement work
being done by member organizations.
SEED PRODUCTION 271
The active membership may consist of any national, state, or
provincial organization carrying on activities in the interest of
improvement of field crops and seeds. At present, between 30
and 35 state crop-improvement associations in the United States
and two Canadian associations belong to the international asso-
ciation. This association has held annual meetings since 1919.
Through the work of its various committees, standards for seed
certification and registration have been developed and applied
in the same general manner by the various certification agencies.
The annual meetings of workers interested primarily in seed
certification and registration have aided in the development of
uniform terminology and in the improvement of methods of seed
certification.
Description of Seed Classes. There arc three general classes
of seed that are recognized by state associations and by the
Canadian Seed Growers' Association, although the terminology,
as used, varies some from one association to another. The
development of more uniform terminology is desirable. The
following definitions serve to point out three types of seed.
1. Foundation stock seed is seed that has descended from a
selection of recorded origin, under the direct control of the original
breeder or of a delegated representative of the state crop improve-
ment association or that is under the control of a state or federal
agricultural experiment station. In many states, such seed is
registered by the seed-certification official of the state crop-
improvement association as "foundation-stock seed."
2. Registered seed is seed of a variety or strain that is the
multiplied progeny of foundation-stock seed and that traces
directly to it. Both registered and foundation-stock seed must
comply with standards of purity and quality laid down by
the state crop-improvement association or other certifying
agency.
3. Certified seed is seed of a variety or strain recommended by
the state agricultural experiment station that has certain required
standards of purity and quality. In certain cases, certified seed
may not meet all the standards required for registered seed.
At least two inspections are necessary in the production of
registered seed, a field inspection during the growing season to
check up on purity of type, admixtures qf other varieties or other
crops, freedom from noxious weeds, etc. If the crop passes this
inspection, a laboratory analysis of a representative sample of
272 METHODS OF PLANT BREEDING
seed is made after harvest and after the seed has been processed.
This inspection often is carried out through the help of the official
state seed laboratory.
The Minnesota Plan for Certain Crops. As an example of
methods, a brief description is given for seed registration of farm
crops in Minnesota. Two classes of registered seed are produced
by members of the Minnesota Crop Improvement Association
called Registered No. 1 and Registered No. 2. A grower pro-
duces registered seed by the plan outlined as follows:
1. Plant registered seed. Obtain seed that has passed the
association requirements and that com.es labeled with the blue
or red tag. Only varieties recommended by the Minnesota
Experiment Station or the board of directors of the Minnesota
Crop Improvement Association are eligible.
2. Plant this seed on clean, well-prepared ground. In the
case of cross-pollinated crops, the field should be 40 rods removed
from any other crops of the same kind.
3. Apply for field inspection before June 15 on crops other than
open-pollinated corn and alfalfa. For these crops, apply before
August 1.
4. Return the application blank, completely filled out, that was
sent you from the office, along with the dues. Where seed was
purchased, also send labels that came with the seed.
After the field and laboratory inspection has been completed,
the certifying official issues blue or red tags for Registered No. 1
or No. 2 seed or rejects the seed when the quality is not up to the
standards.
The requirements in order to pass the field inspection are
summarized in the following statements :
1. Fields will be rejected if plants of field bindweed, leafy
spurge, or other noxious weeds, the seeds of which are extremely
difficult of separation, are found in the field. If the noxious-
weed seeds can be easily separated from the seed crop, the presence
of plants of these noxious weeds will not be sufficient cause for
rejection of the field.
2. More than a mere trace of other crop plants or plants of
other varieties will be cause for rejection of the field. Mixtures
that may cause rejection include (a) sweet clover in alfalfa,
(6) durum wheat in spring wheat, (c) winter wheat in winter rye,
or vice versa, and (d) timothy in alsike clover.
SEED PRODUCTION
273
3. Instructions are given at the time of inspection regarding
roguing, harvesting, and other matters pertaining to the handling
of the seed crop.
For hybrid corn, certain isolation requirements are necessary.
As a general practice, the seed plot should be 40 rods from other
corn. This isolation may be provided by natural barriers, actual
distance, male border rows, or a combination of these methods.
If natural barriers are used, permission must be obtained in writ-
ing from the registration official before planting, in addition to
the regular field inspection. Border rows of the male parent may
be used to reduce the actual distance required if the corn inter-
fering with the isolation of the hybrid plot is of the same color
as the female parent.
The following applies only to detasseling plots of 5 acres or
less:
Number of Border Rows Actual Distance Female Parent Must Be
of Male Parent Removed from Other Corn, Rods
1
2
3
4
5
6
7
8
9
10
11
12
40
35
30
25
20
15
10
This minimum distance of 10 rods may be reduced on larger
fields according to the following plan, provided corn of another
color is' not involved :
Male border rows
needed
Actual distance female parent
is removed from other corn, rods
Size of field,
acres
13
9
10
13
8
15
14
7
20
14
6
25
15
5
30
15
4
35
16
3
40
274
METHODS OF PLANT BREEDING
The plot shall be detasseled according to instructions furnished
for the production of the hybrid. Three inspections during the
detasseling period will be made to determine if the work is satis-
factorily carried out.
Laboratory requirements for Registered No. 1 and Registered
No. 2, the two grades of seed recognized by the Minnesota Crop
Improvement Association, are given in Table 31 for small grains,
alfalfa, and hybrid com.
TABLE 31. REQUIREMENTS FOR REGISTERED No. 1 AND REGISTERED No. 2
SEED FOR SMALL GRAINS, ALFALFA, AND HYBRID CORN
Maximum allowance,
weed seeds
Maxi-
Labora-
Sec-
m u 111
allow-
Inert
mat-
Ger-
mina-
Grade
Tug
tory
ond-
ance,
ter,
tion,
purity,
per cent
Noxi-
ary
noxi-
Other,
per
01 op
seeds,
per
cent
per
cent
ous*
per
ous, t
cent
cent
pei-
cent
Small Grams:
Registered No. 1
Bluo
99-100
None
01
10
0.10
1.00
90-100
Registered No. 2 .
Red
98-99
Noue
05
0.15
0.30
2.00
70-89
Alfalfa:
Registeied No. 1 . .
Blue
99 3 100
None
01
0.10
20
70
90-100
Registered No. 2
Red
98.5-99 3
None
05
15
0.50
1 50
70-89
Germina-
Grad-
Mois-
Dam-
ture,
age,
tion
ing
per cent
per cent
Hybrid Corn:
Registered No. 1
Blue
93-100
90-100
14
0.5
Registered No. 2. , ....
Red
75-92
75-89
15
1 5
* Primary noxious weeds: Canada thistle, perennial sow thistle, quack grass, dodder,
buckthorn, oxeye daisy, field bindweed, horse nettle, leafy spurge, Austrian field cress, false
flax (in flax), and perennial pepper grass.
t Secondary noxious weeds: Wild mustard, French weed, wild oats, wild vetch, sheep
sorrel, hedge bindweed, night-flowering catchfly, white cockle, and dragonhead mint.
No primary noxious-weed seeds are allowed in either class of
registered seed. The requirements for Registered No. 1 are
considerably higher than for Registered No. 2.
SEED PRODUCTION
275
The blue tag given for Registered No. 1 may be illustrated as
follows :
REGISTERED No. 1 SEED
THIS SHIPMENT CONTAINS..
VARIETY NAME
<*> PURITY
% GERM.
TESTED
MONTH
YEAR
WEED SEEDS
5 IS SAFE SEED GROW
5 No Primary Noxiou* Weeds p o _
REGISTERED No. 1 SEED
GROWERS CERTIFICATE
I hereby certify that the seed contained
in this sack was produced by me in 19
in accordance with the rules of the Minn-
esota Crop Improvement Association. That
it is of the kind, variety, amount and germ-
ination as stated on the reverse side of this
tag.
That it conforms to the standard of pur-
ity, grade and cleanliness for Registered
No. 1 Seed.
THE REGISTRATION NO IS_
MINOR IMPURITIES ARE
276
METHODS OF PLANT BREEDING
Tags used in registering hybrid corn seed are similar to those
used for small grains. The material carried on the tag is different.
No. 1 REGISTERED
HYBRID SEED CORN
From_
Ship to_
IS SAFE SEED
Of Known Inheritance
REGISTERED HYBRID SEED
CORN GROWERS CERTIFICATE
I hereby certify that the seed contained
in this sack was produced in 19.. in ac-
cordance with the rules of the Minnesota
Crop Improvement Association and that it
conforms to the standards stated below:
Minhybrid
Type of Crosa_
Approx. Days Maturity.
Registration No
Grade
Germination
Purity %
Moisture %
County Grown_
Year Grown
Date of Test_
(Void unless completely filled out)
SEED PRODUCTION 277
In addition to the blue- and red-tag grades of hybrid corn seed,
the Minnesota Crop Improvement Association certifies hybrid
corn seed for seed companies if they comply with requirements.
A green tag is furnished bearing the same information as required
for hybrids produced by growers of experiment-station hybrids.
The tag carries the following statement:
The seed in this container is certified on the affidavit of the producer
filed with the Minnesota Crop Improvement Association that the
parent lines used in the commercial crossing plot are of the same breeding
and purity as those parental lines used in the hybrid tested in the official
yield test. Regular inspections were made of the commercial crossing
plot for isolation, detasseling and purity. Representative samples of
the seed, as prepared for market, were inspected in the laboratory for
moisture, germination, grading and physical appearance as required
in the rules for certification.
Hybrids produced by seed companies that have been in
official yield trials and that have given satisfactory performance
are eligible for certification.
Seed (Tuber) Certification for Potatoes. At the present time,
there are 22 states actively engaged in the production of certified
seed. In 10 states, certification is under the direction of the
college of agriculture. These are Colorado, Idaho, Louisiana,
Maryland, Michigan, Montana, New York, Oregon, Wisconsin,
and Wyoming. In 9 states, the work is under the supervision
of the state department of agriculture. These are California,
Maine, Minnesota, New Jersey, North Dakota, Pennsylvania,
Tennessee, Vermont, and Washington. In Nebraska, Utah, and
South Dakota, growers' organizations have charge of certifica-
tion; in Canada, the work is under the supervision of the Domin-
ion Department of Agriculture, at Ottawa.
The objects of seed certification of potatoes include:
1. The production of high-grade seed potatoes that are rela-
tively free from diseases and varietal mixtures and that are
well graded.
2. Increased yield and better quality that follow the use of good
seed stock, relatively free from disease.
3. More satisfactory prices of seed stock to careful growers of
certified seed.
4. Better methods of production of tubers used for seed.
278
METHODS OF PLANT BREEDING
Sufficient fees are charged for inspection so that the cost of the
work is paid for largely by the producer of certified seed.
The principal steps in potato-seed certification, as carried out
in Maine, are briefly summarized.
Potatoes eligible for certification should be grown on land that
was not in potatoes the previous year and on fields isolated by
250 ft. from other potatoes. It is recommended that certified
seed be used to plant the field and that such seed be disinfected
with corrosive sublimate. It is required that the crop be well
cared for and be kept reasonably free from weeds and from
injury by insects. It is required also that the field be sprayed
with Bordeaux mixture to control late blight.
Two field inspections of the crop are made; the tolerances
allowed for various diseases and varietal mixtures are given in
the summary.
Tolerances allowed for
diseases and varietal mixtures
First
inspection,
per cent
Second
inspection,
per cent
Leaf roll
Mosaic
2
3
1
2
Spindle tuber
2
2
Yellow dwarf
Total virus diseases
Blackleg
5
5
2
0.5
3
1
Wilt
2
1
Total of all diseases
6
4
Giant hills ....
1
Varietal mixtures
1
25
It is expected that the grower will remove diseased hills or
varietal mixtures after each inspection.
At shipping time, a third inspection must be made. Maine-
certified seed potatoes shall be equal or exceed II. S. Grade No. 1
to be eligible for certification.
SEED PRODUCTION 279
The blue tag used to designate certified seed is illustrated here.
MAINE
CERTIFIED SEED POTATOES
CROP OF 1941
Crop inspected twice in field and tuber inspection at time of
shipping. CARL R. SMITH, Commissioner of Agriculture
Date
Maine Department of Agriculture
DIVISION OF PLANT INDUSTRY
Variety
Tha MMd in thu packagr i from held* inpec'd and patted by
the Maine Department of Agriculture -.
CROP OF 1941
Grower^
600002-
Final inspection
made by
Address
CHAPTER XVIII
SOME COMMONLY USED MEASURES OF TYPE AND
VARIABILITY
"~-N
Statistics are being used extensively in the reduction of data
and interpretation of results from plant-breeding experiments.
Whenever a large number of observations are obtained, it will
be difficult to grasp the full importance of these observations
because of their number. Consequently, the individual observa-
tions are replaced by a few statistics that convey all or most of
the information available from the experiment in a form readily
comprehended.
One of the commonest uses of statistics for the plant breeder
is in their application to field trials where a considerable number
of varieties are grown under comparable test. In such trials,
it is necessary to determine the averages of yield and other
characters and to estimate the significance of differences As
usually carried out, the first test is to determine whether there
is a significant difference in the performance of any of the varie-
ties. If the statistical method used indicates that all varieties
have the same performance, within the limits of the accuracy
of the study, no further comparisons are worth while. If there
is a significant difference in performance, i.e., if the odds are
rather great that the difference in performance would not occur
by chance alone, the next step is to compare individual varieties.
For the plant breeder, this often will consist of a comparison
between new selections that have been placed under test recently
and a standard variety that has been shown previously to be the
most desirable variety available.
Before the investigator can make these and other comparisons
of a similar nature, it is necessary to learn the meaning of certain
statistical terms and the method of their calculation.
DEFINITION OF STATISTICAL CONSTANTS
The commonest statistics are the mean and mode as measures
of type and standard error, standard deviation, and variance ac
measures of variability.
SOME COMMONLY USED STATISTICS 281
The mean, or arithmetic average, is the sum of the measures or
observations divided by their number.
The mode is the class of greatest frequency in a series.
The standard error is a measure of variability in terms of the
units of measurement. The reliability of a particular statistic
is determined by its standard error. The smaller the standard
error in relation to the magnitude of the statistic the greater the
confidence that may be placed in the significance of that statistic.
The standard deviation is similar to standard error except that
it frequently refers to the infinite population rather than to any
sample drawn from that population.
The variance is the square of the standard deviation or stand-
ard error.
The coefficient of variability is a measure of variability
expressed in percentage of the moan, making it possible to com-
pare the relative variability of two populations with widely
different means.
CALCULATION OF MEAN, STANDARD ERROR, VARIANCE, AND
COEFFICIENT OF VARIABILITY
The calculation of these statistics will be illustrated by using
data given by Mercer and Hall (1911) on yield of 500 small plots
of the same variety of wheat harvested from one field. In Table
32 is given the frequency of occurrence of plots where the yields
have been grouped into classes of 0.2 Ib. per plot.
In the calculations that will be illustrated, S means summation,
/ is the frequency or number of plots having a certain yield, x is
the class-center value, and N is the total number of plots.
The formula for the calculation of the mean is as follows:
Mean yield = x = S(fx)/N.
In the problem: S(fx) = S[(4 X 2.8) + (15 X 3.0) -
+ (4 X 5.2)] = 1974.6. To obtain the mean yield, this value is
divided by N, where N is the total number of plots. Numeri-
cally, this would be 1974.6 -*- 500 = 3.9492 Ib. per plot.
The mode is the class with greatest frequency. In this
problem, the modal class is 4.0 Ib. Plus and minus deviations
from the modal class often are similar in their frequencies; i.e.,
the distribution is frequently symmetrical about the modal class.
In pure-line material, these deviations are the result of the inter-
action of favorable and unfavorable environmental influences,
282
METHODS OF PLANT BREEDING
the number of individuals or plots where all conditions are favor-
able, or all unfavorable, being much smaller than those with part
favorable and part unfavorable. In segregating lines, there may
be also variation due to heritable causes, and in some cases a
frequency distribution may show a bimodal curve.
TABLE 32. FREQUENCY OF PLOTS WITH YIELDS GROUPED INTO 0.2-LB.
INTERVALS
Class center of
yield, x
Number of plots,
/
/*
fx*
2 8
4
11.2
31.36
3
15
45
135 00
3 2
20
64
204 80
3.4
47
159 8
543 32
3 6
63
226.8
816 48
3.8
78
296 4
1126 32
4
88
352
1408.00
4.2
69
289 8
1217.16
4 4
59
259 6
1142.24
4.6
34
156.4
719.44
4.8
11
52.8
253 44
5
8
40.0
200 00
5 2
4
20.8
108.16
Total
500
1974^6
7905.72
The common measures of variation are the standard error and
variance, the latter being the square of the standard error.
The standard error is given by 5
fry / >
&(}
=v~
- S(fx)x
where
N - 1
S, /, x, and N have the same designations as above. From the
- . . U1 .,. ,,, /7905.72 - (1974.6) (3.9492)
foregoing table, this would be * ^ =
0.464 Ib. The foregoing formula also may be expressed as
r _i This is the standard error of a single determina-
tion. In practice, the value of the mean x used in the correction
factor S(fx)x must be calculated with sufficient accuracy so that,
when multiplied by the total, the product is accurate to the
place desired, Usually it is more convenient to calculate the
correction factor in the form [S(fx)]*/N,
SOME COMMONLY USED STATISTICS
283
In Fig. 36 is given a histogram of the frequency of plots with
different yields. In the same figure is superimposed the normal
frequency distribution. This smooth curve is an estimate of
the distribution of the infinite population from which these 500
plots are considered a sample.
90
"
/I
T*S
\
/i>
/
\
y
/
\
w go
/
\
"a.
^ ..
/
\
V
45
Q>
J3
\
S.
330
/
\
|C
/
v
y
\_
J=^
^
T^^
~4a -3cr ~2cr -Id M -Her +Zcr +3cr +4or
FIG. 36. Yield in pounds per plot with ordinates drawn at 1, 2, 3, and 4 times
the standard deviation.
In the illustration given, and by the use of tables such as
Sheppard's (Pearson 1924), it is possible to determine the per-
centage of the area under the normal curve cut off by erecting
perpendiculars from the base line to the curve at distances of
1, 2, or 3 standard deviations (l<r, 2<r, or 3cr, as illustrated in Fig.
36) from the mean, or M, of the curve, where <r is the standard
deviation. From such tables, we find that the area between the
two lines erected at plus and minus 1 times the standard devia-
tion will be 68.27 per cent of the total area. It can be said, then,
that the probability of an observation falling within Icr will
be given by P = .6827. The probability of an observation fall-
ing outside Iff will be 1.0000 - .6827 = .3173. In like manner,
the probability of an observation falling within 2cr will be
.9545, and within 3<r it rises to .9973. In the problem con-
sidered in Table 32, the mean was 3.95, and the standard error
was 0.464. One can say, therefore, that the probability of an
observation falling within the limits set up by the mean 2(r, or
3.95 + 2(0.464) = 4.88 and 3.95 - 2(0.464) = 3.02, or between a
yield of 3.02 and 4.88 Ib, is 0.9545. The probability of
284 METHODS OF PLANT BREEDING
the yield of a plot, selected at random, falling outside the limits
4.88 and 3.02 would be .0455. Stated in another way, it may be
said that the chances of an observation falling within these two
limits will be .9545:. 0455, or approximately 21:1. There is
1 chance in 22 of a plot, selected at random, falling outside the
limits 4.88 and 3.02. The chances of an observation exceeding
4.88 would be 43: 1. Stated in terms of probability, it would be
.0455 2 = .02275, or in a little over 2 per cent of the cases by
chance alone would a plot selected at random yield over 4.881b.
The standard error of a mean is given by s/\/N where s =
standard error of a single determination and N = the number
of observations from which the mean is determined. It is
obvious that the standard error of a mean must be smaller than
the standard error of a single determination, since there will
be less variation among means than among single observations.
The coefficient of variability (C.V.) is a relative measure
of variability in percentage. C.V. = (s X 100)/.x, where s and x
are the standard error and mean for the sample. It is of value
in comparing the variability of populations with different means
or differing in units of measurement.
In many cases, the investigator will be interested in a com-
parison between the means of two varieties or treatments or in a
comparison of a selection with the standard. It is essential to
place these comparisons on the basis of probabilities. The usual
procedure is to compare the difference with its error and deter-
mine the probability that a difference as great as or greater than
that observed could be due to chance alone.
The standard error of a difference (S^H ) is given by the formula
~
+ si ~ 2r a bS a Sb, where a arid b represent the two treat-
ments being compared and r is the correlation between sepa-
rate measurements of the quantities. When r = 0, the formula
becomes \A + s f. When s a = s b and r = 0, s d if f . becomes
s \/2. In this and in other similar problems, the significance
of a difference is determined by comparing it with the standard
error of the difference. These problems will be taken up in
detail in later chapters.
CORRELATION COEFFICIENT
The coefficient of correlation r is used as a measure of the degree
of association between two characters worked with at the same
SOME COMMONLY USED STATISTICS 285
time. Perfect positive correlation is +1, and perfect negative
correlation is 1; no correlation is given by r = 0; intermediate
values denote association of an intermediate degree. A con-
venient working formula is
_ S(xy) - S(x)S(y)/N
VS(x) - [S(x)]*/N VW) - [S(y)]*/N
where x represents the measures of one variable or character and
y the other and all other letters have the same meaning as before.
The caleulaton of simple, partial, and multiple correlation coeffi-
cients will be given in a subsequent chapter.
COMPARISON OF DIFFERENCES BY THE t TEST
The t test provides the usual method for testing the signifi-
cance of the difference between two means. Such problems
are common in plant breeding. The statistic t is defined as a
difference expressed in terms of the standard error of the differ-
ence. If we have two varieties, or treatments, from which the
means are different, we shall wish to know in what proportion
of the cases a difference as great as or greater than that observed
can be expected to occur as a result of deviations due to random
sampling.
Tests of this type fall into two classes: (1) when the samples are
paired and (2) when the samples from one of the two varieties or
treatments are not paired with those of the other. Both tests
will be illustrated with data obtained in Minnesota from strip
plantings of two varieties of wheat. Seed of the varieties
Thatcher and Marquillo was sown in adjacent single strips of
one drill width in each of many fields in the state. The purpose
of the test was partly demonstration al and was partly to obtain
comparative yields on many different farms. The yield was
determined from comparable samples from each variety in each
strip. In Table 33 are given the yields of these two varieties
for 12 of the many farms on which tests were made and the sums
and differences of the two varieties.
Since the two varieties were grown in paired plots, this fact
will be utilized in the statistical analysis of the differences.
In this problem, differences in yield between Thatcher and
Marquillo were determined for ea^ch of the 12 comparisons, and
286
METHODS OF PLANT BREEDING
the mean difference, called x, divided by the standard error of
this difference to obtain the value of t.
TABLE 33. YIELDS OF THATCHER AND MARQUILLO WHEAT TESTED
IN COUNTY DEMONSTRATION TRIALS IN MINNESOTA IN 1935
Farm number
Yield, bu
. per acre
QIITYI
and county
Thatcher
Marquillo
1 . Roseau
2. Mahnornen
24.4
27 9
17.5
15 1
41 9
43
6.9
12.8
3. Traverse
28,2
21.6
49 8
6.6
4. Bigstone
5 Stevens
19 8
23 1
18.2
21 6
38
44 7
1.6
1 5
6. Stevens
22 9
13.7
36 6
9.2
7. Pope
25.6
24.8
50.4
.8
8. Kandiyohi
28.7
27.8
56.5
.9
9 Kandiyohi
26 2
25 2
51 4
1
10. Kandiyohi
25 7
19.2
44 9
6.5
11. Renville
37.0
34
71
3.0
12. Yellow Medicine
31 .5
25 2
56 7
6 3
Sum
321.0
263 9
584.9
57.1
The calculated mean difference was obtained by dividing
S(x) by N, or 57.1 by 12, which gives a mean value of 4.76 bu.
In other words, Thatcher yielded, as an average of the 12 trials,
4.76 bu. per acre more than Marquillo.
The standard error of a difference was obtained by the same
formula as that presented previously, except that / = I , where
= j.
'S(x*) - [S(x)]*/N
N - I
From Table 33, S(x*) was computed by squaring each of
the 12 differences to give 437.85, and the value of s becomes
'437.85 -
~
Qn , , , . .,
= 3.89. The standard error of the
mean difference is 3.89/\/12 or 1.12.
The statistic t (the mean difference divided by its standard
\ i_ x 4.76 , OF
error) is given by t = r- r~ = 4.25.
l.lz
In a determination of the significance of this difference, it is
necessary to introduce the concept of degrees of freedom. That
SOME COMMONLY USED STATISTICS 287
term is used in the sense of independent comparisons. In
calculating the standard error of a difference for the 12 compari-
sons, or differences in yield between Thatcher and Marquillo,
only N 1 deviations can vary, one being fixed by the sample
mean. The degrees of freedom are 1 less than the number of
comparisons, or 11 in this case.
Referring to Appendix Table I for 11 degrees of freedom, we
find the values of t for the 5 and 1 per cent points to be 2.20 and
3.11, respectively. Values of t as great as these give odds of
19:1 and 99:1, respectively, against a difference as great as this
occurring by chance alone. In this problem, with a value of
t = 4.25, the chances are much greater than 99:1 that the
difference is not due to chance.
In some cases, it may be desirable to compare the yields of
two varieties when they are not grown in paired comparisons.
Fisher (1938) has given a method where the value of t is calculated
by comparing the mean difference in yield of the two varieties
with its standard error, where x\ and x% are the mean yields,
s = the standard error, and Ni and N% are the number of plots
of each variety. Then
where s 2
(JVl - i) + (N* -
For such comparisons of means, the number of plots of each
variety need not be equal. Calculating the sums of squares of the
12 yields of Thatcher and Marquillo, we obtain 219.55 and 365.31,
respectively. Then
/219.55 + 365.31 __ K 1A
o.lo
5.16 \ 24 " 2.11
The degrees of freedom for comparing these two varieties in
unpaired comparisons are 2 less than the sum of the number of
trials of the two varieties, or 22 in this case. Entering the i
table for t ~ 2.26 and 22 degrees of freedom, it is seen that the
observed value of t lies between P = .05 and P = .01. The
288 METHODS OF PLANT BREEDING
chances that a difference as large as the observed would occur
through random sampling would be less than 5 in 100 and more
than 1 in 100.
It sometimes happens that in comparing two means based on
the same number of observations, the data may be analyzed
according to cither method illustrated above. If either method
gives a significant value of f, its testimony should not be ignored.
With the paired relationship, the degrees of freedom will be one-
half as large as when the data are not paired. This will result
in a larger difference being required to reach the minimum level
of significance because of the reduced number of degrees of
freedom. If the correlation between paired plots is sufficiently
high, the standard error of the difference will be reduced suffi-
ciently so that the minimum level of significance is smaller in
spite of the reduced degrees of freedom.
CHAPTER XIX
FIELD-PLOT TECHNIC
After making the initial selections from promising material,
the final test in plant-breeding studies will consist of comparable
trials in which the selected material will be compared with
standard varieties. Although special technics may be adopted
for studying such characters as winter hardiness and drought
resistance, disease resistance, and other special qualities, it will
be necessary in most cases to make comparable yield trials under
actual field conditions. In these cases, the field becomes the
experimental laboratory. The usual method consists of the use
of small plots, adequately replicated and dealt with in such a
manner as will give a reliable index of comparable yielding
ability under actual farm conditions. In order to obtain the
desired results, it has been found necessary to handle the experi-
mental field, insofar as possible, in a manner that will approach
the practices in use by the better farmers, i.e., to follow estab-
lished principles of farm management. Some of the important
considerations may be summarized:
1. As far as possible, the soil and climatic conditions of the
experimental field should be similar to those under which the
crop will be grown by farmers.
2. A system of crop rotation should be followed that ap-
proaches, as closely as feasible, that used by the better farmers.
3. A bulk crop sown after the experimental trial aids in keeping
the soil in a uniform state of fertility.
4. Competition between varieties and strains in the experi-
mental trial must be eliminated or its effects controlled by
randomization or by grouping varieties of like nature.
5. Satisfactory methods must be devised for handling the
experimental plots, including weighing or counting the seed for
planting, sowing, cultivating, harvesting, and threshing the crop.
6. Replication and the calculation of a standard error of the
experiment aid in furnishing a basis for reliable conclusions, and
289
290 METHODS OF PLANT BREEDING
a well-designed experiment helps in controlling the effect of soil
heterogeneity.
Some of these points seem self-evident; others need to be
explained in greater detail. Each experiment must be planned on
the basis of the information desired. All that will be attempted
is the formulation of principles of wide application.
In order to determine the adaptation of varieties to actual farm
conditions, it is necessary to make the field trials in various
regions. For this purpose, it has become a standard practice
to develop branch stations or test on selected farms in representa-
tive regions in order to test the new strains under those conditions
to which they will be exposed after introduction to the farmer.
These experimental fields should be operated according to
approved farm-management practices. Among these practices,
the importance of an adequate system of conserving soil fertility
will be generally appreciated. Crop rotation will be desirable
in many cases.
CROP ROTATION FOR EXPERIMENTAL FIELDS
The system of rotation used should be similar to that recom-
mended as a desirable farm practice. Several such rotations
may be illustrated, although they are representative only of
desirable practices for certain specific types of farming.
For the corn-yield trials at Minnesota, a 3-year rotation is
practiced. In this case, one-third of the land is used in the yield
trials. The rotation is as follows :
1. Corn-yield trials. Farm manure is applied, and super-
phosphate is added at the rate of 100 Ib. per acre.
2. Small grain follows corn, this field being used to increase
seed of a recommended variety.
3. Sweet clover is sown with the small grain and used for
pasture or hay the third year.
Rotations have been developed for the final trials of spring and
winter wheat made in )^o~&cre plots that have proved satis-
factory and that are similar to systems in use by Minnesota
farmers. These need not be given in great detail.
Spring Wheat Winter Wheat
1. Yield trial 1. Yield trial
2. Clover hay 2. Bulk corn for silage
3. Bulk corn for silage 3. Oats and peas for hay
FIELD-PLOT TECH NIC 291
Manure is added to the silage corn, which aids in maintaining
soil fertility. Clover is planted with the spring wheat, and the
first crop of clover is used for hay the following year, the second
crop being turned under. The oats and peas for hay are har-
vested sufficiently early to permit adequate preparation of the
land for the winter wheat. A part of this winter-wheat series
is used for the winter- wheat breeding nursery.
At Cornell, the following general rotation for the experimental
trial of cultivated crops has been used. The plan is as follows:
1. Soybeans as green manure, turned under.
2. Silage-corn-yield trials, manured, with addition of super-
phosphate, or yield trials of cabbage.
3. Yield trials of soybeans, field beans, sunflowers, or corn.
Such a rotation results in high fertility, and the yields are high
if the seasonal conditions are satisfactory. This plan is followed
because high yields are expected by New York farmers who grow
silage corn and the other cultivated crops worked with.
In the rotations for these cultivated crops, as well as in the
small grain trials at Cornell, variety trials may be made on the
same fields for 2 consecutive years, and then a legume is turned
under the third season. The small-grain trials are in a 3-year
rotation, consisting of the following: (1) rod-row trials of oats,
(2) rod-row trials of wheat, (3) clover cut off once and then
plowed under.
These rotations will serve to illustrate methods that have
proved fairly satisfactory. Although it is better to have a sepa-
rate rotation for each yield trial, where the yield comparison of a
crop occupies the position taken by that crop in a desirable crop
sequence for the locality, it is not always possible to follow this
plan because of insufficient cropland.
SOIL HETEROGENEITY
One of the difficulties encountered in field experiments is due
to the fact that uniform soil conditions seldom exist, if ever, even
over small portions of a field. Soil heterogeneity, as measured
by yield of crops grown on a field and harvested as small plots,
may be due to topography of the field, soil moisture, variation
in fertility, or previous cropping practice.
In 1915, J. Arthur Harris proposed a criterion for measuring
soil variation that he called a coefficient of soil heterogeneity.
292 METHODS OF PLANT BREEDING
Five years later Harris (1920) reported the results of tests made
on published data involving a wide variety of crops and charac-
ters from experiments conducted over the entire world and
demonstrated clearly that soil heterogeneity is practically
universal. In concluding his paper, Harris stated, "The demon-
stration that the fields upon which the plot tests have been carried
out in the past arc practically without exception so heterogene-
ous as to influence profoundly the yields of the plots emphasizes
the necessity for greater care in agronomic technic and more
extensive use of the statistical method in the analysis of the data
of plot trials if they are to ho of value in the solution of agri-
cultural problems. " The many studies conducted since that
time have amply substantiated these conclusions.
Uniformity trials, or blank tests, have been used extensively
in studying the nature and extent of soil heterogeneity. In
such uniformity trials, the field is planted to a single variety
and harvested as small plots. The entire field is planted at
the 'same rate of seeding, and cultural practice is the same
bver the entire area. The unit plots harvested can then be
grouped to form plots of varying size and shape, the only vari-
able being size or shape of the plots, Cochran (1937) published
a catalogue of uniformity-trial data, listing 191 uniformity
trials with field experiments, the data from 135 having been
published.
The nature of soil heterogeneity may be demonstrated in
graphical form by means of contour maps drawn from data
obtained in uniformity trials. An example of such a contour
map, drawn from data on yield of sugar beets in a uniformity
trial, was given by Immer and Raleigh (1933) and is reproduced
in Fig. 37.
In this study, the yields of six-row plots, each 2 rods long, were
used. Points deviating by 15, 10, 5, 0, +5, and +10 per
cent from the mean yield were interpolated between the centers
of the plots and the contour map drawn by connecting these
points. Figure 37 shows in a graphic way that fields that may
appear to be very uniform are rather heterogeneous from the
standpoint of productivity, as measured by yield on small areas.
Such contour maps demonstrate graphically that soil variability
is, to a certain extent, regular over small areas. There is a sort
of "regular irregularity 77 to the fertility contours. The use of
FIELD-PLOT TECHNIC
298
different sizes or shapes of plots would change the contour map,
but the general characteristics would remain the same.
The extent of soil heterogeneity may be measured by determin-
ing the degree of correlation of yields of near-by plots. Hayes
and Garber (1927) presented data giving the correlation coeffi-
456
Block no.
10
-20 -15 -10 -5 45 410 415
Scoile of shades
FIG. 37. Contour map of weight of sugar-beet roots in a uniformity trial from
100 six-row plots each 2 rods long; contour lines drawn through the points
deviating by 15, 10, 5, 0, +5, and +10 per cent from the mean weight.
cient between adjacent rod rows within tests with oats, spring
wheat, and winter wheat and between rod rows separated by one
or more plots. These data are reproduced in Table 34.
It is evident that the correlation is greatest between adjacent
plots and decreases as the distance between the plots is increased.
However, there was a sensible correlation between the yield of
rod-row plots separated by as much as 10 rows.
Harris (1920), using the intraclass correlation coefficient, made
an extensive study of soil heterogeneity, using data from uni-
formity trials obtained- by numerous investigators. The extent
to which contiguous plots resemble each other was measured in
terms of intraclass correlation, the larger the coefficient the
greater the degree of soil heterogeneity. The results are given
in Table 35.
The data given in Table 35 are only a small part of those avail-
able but are presented to emphasize the usual extent of soil
heterogeneity in plot studies.
The method of calculating simple correlation coefficients will
be taken up in Chap. XX. The method of intraclass correlation
will be illustrated here. Harris calculated the intraclass cor-
294
METHODS OF PLANT BREEDING
TABLE 34. CORRELATION OF PERCENTAGE YIELDING ABILITY IN NEAR-BY
PLOTS OF OATS, SPRING WHEAT, AND WINTER WHEAT, 1924
Crop
Correlation of
Correlation
coefficient
I
Oat-rod rows
Adjacent plots
Separated by 1
Separated by 2
.572 .025
.490 029
.407 .034
Spring- wheat rod rows
Separated by 3
Separated by 4
Separated by 10
Adjacent plots
Separated by 1
Separated by 2
.412 .035
.264 .041
.275 .057
.618 .023
.518 .028
.454 .030
Winter-wheat rod rows <
Separated by 3
Separated by 4
Separated by 10
Adjacent rows
Separated by 1
.383 .034
.449 .034
.429 + .060
.552 .068
.293 + .028
Separated by 4
- .114 .118
TABLE 35. CORRELATION COEFFICIENTS PRESENTED BY HARRIS THAT
EXPRESS THE EXTENT OF SOIL HETEROGENEITY IN DIFFERENT
LOCALITIES AND WITH DIFFERENT CROPS
Crop
Character
Size of plot
Investigator
Correlation
coefficient
Wheat
Yield, grain
5.5 by 5.5 ft.
Montgomery, Nebr.
.603 .029
Wheat ....
Nitrogen content
5 . 5 by 5 . 5 ft.
Ivloiitgomery Nebr.
.115 .044
Oats
Yield, grain
i/ n acre
Kiesselbach Nebr,
.495 .035
Mangels. . . .
Yield, roots
To wvi c
}-2oo acre
Mercer and Hall, England
.346 .042
(Rothamsted)
Mangels ....
Yield, leaves
^oo acre
Mercer and Hall, England
.466 .037
(Rothamsted)
Potatoes
Yield
Rows, 72 ft., 7
Lyon
.311 .043
in. long
Cora . ,
Yield, grain
\4 rt acre
Smith, 111 (1895)
.830 .019
Alfalfa
Yield, hay
710 **v*c
1913, first cutting
0.085 acre
Scofield, Huntley Experi-
.407 .059
ment Farm, Montana
1913, second cut-
0.085 acre
.343 .062
ting
1914, first cutting
. 085 acre
.602 .045
1914, second cut-
. 085 acre
.657 .040
ting
FIELD-PLOT TECHNIC
295
relation coefficient (which he designated a coefficient of soil
heterogeneity) from the formula appropriate for analysis of a
symmetrical correlation table. The manner of calculating the
intraclass correlation as given by Fisher (1938) is illustrated
from the simple illustration given below, assuming data from 16
plots in a field divided into four blocks of four each.
4
5
4
3
5
6
5
4
6
6
5
5
5
7
6
4
The sum of the 16 plot yields is 80 and the mean is 80 -f- 16 = 5.
The total sum of squares is calculated from S(x*) S(x)x J where
x = individual plot yields, x = mean yield, and S = summation,
or 416 - (80) (5) = 16. The sums of the yields of the four blocks
of four plots each are 20, 16, 24, and 20. Squaring these four
sums, dividing by the number of plots in each sum, and subtract-
ing the correction factor S(x)x gives the sum of squares between
blocks. Numerically, this is 163 % 400 = 8. The sum of
squares within blocks is obtained by subtraction. The degrees
of freedom for total and block variation will be 1 less than the
number of total plots or blocks. An analysis of variance is
given in Table 36,
TABLE 36. ANALYSIS OF VARIANCE
Variation
Degrees of
freedom
Sum of
squares
Mean
square
F
Between blocks
3
8
2.667
4.00
Within blocks
12
8
667
Total
15
16
The mean square is obtained by dividing the sums of squares
by the respective degrees of freedom. F is the mean square for
blocks divided by the mean square for variation within blocks
296 METHODS OF PLANT BREEDING
If the mean square within blocks is set equal to B and the mean
square between blocks equal to (kA + -B), where k is the number
of plots per block, the intraclass correlation coefficient will be
given by
A
A + B
In this illustration, B = 0.667, k = 4, and kA + B = 2.667.
Therefore kA = 2.000, and A = 0.500. The intraclass correla-
tion will be %
0.500
0.500 + 0.667
= +.428
The significance of the intraclass correlation coefficient may bo
determined from a comparison of mean squares between and
within blocks. In this case, the mean square between blocks
divided by the mean square within blocks gives a value for F of
4.00. Themeaning of F will be discussed in greater detail in Chap.
XX. An F of 4.00 is in excess of the 5 per cent point (Appendix
Table II) for n\ = 3, and n 2 = 12 degrees of freedom, n\ and n 2
being the degrees of freedom for the larger and smaller mean
squares, respectively. The intraclass correlation therefore may
be judged significant. This coefficient presents in correlation
form the ratio of the variance between and within blocks. It
expresses the average correlation of plots within each group of
adjacent plots that are studied.
It is important to know whether there is a tendency for plots
that produce low yields in one season to produce low yields in
succeeding seasons, etc. The results given by Harris and
Scofield (1920, 1928) indicate a tendency for plots to yield in a
similar manner from year to year, although there were some
exceptions. Garber, Mclllvaine, and Hoover (1926) found the
interarmual correlation between the yields of oat hay on 270 plots
in 1923 and the yield of wheat grain on the same plots in 1924
was +.364. Garber and Hoover (1930) found that the inter-
annual correlation between yields of plots in uniformity trials
and yields of crops in a rotation test on the same field in succeed-
ing years was positive in each test made. The differences in
natural productivity of the plots persisted over a period of several
years,
FIELD-PLOT TECHNIC
297
Summerby (1934) presented the most extensive data on a study
of permanence of difference in yields of crops over a period of
years. The interannual correlation coefficients were mostly
positive, but 13 out of 143 were negative, and 3 of these exceeded
the 5 per cent point. As a result of his study, Summerby con
eluded, "Under the conditions of this experiment the use of
preliminary uniformity trials for the purpose of adjusting yields
of subsequent experiments by regression is only rarely as effective
in increasing precision as is the use of the same amount of land
and labor in replicating the experiment in the year of the
trial."
COMPETITION
Plants growing along the sides and ends of plots frequently are
larger than those in the middle of the plot because of greater
available fertility or water supply. This is true particularly when
the plots are adjacent to uncultivated areas or are surrounded
by alleys. The extent and nature of this border effect is impor-
tant in comparative crop tests.
, Amy and Hayes (1918) and Amy (1921, 1922) studied the
extent of border effect in multiple-row plots. In Table 37 are
presented data obtained by Amy and Hayes from plots seeded
with a grain drill in rows 6 in. apart. Eighteen-in. alleys sur-
rounded the plots with a roadway at each end. The plots were
17 drill rows wide and were trimmed to 132 ft. in length. Each
of the outside border rows was harvested separately, and the
yields compared with those of the 13 central rows. The plants
on the end, to a depth of 1 ft., were cut off.
TABLE 37. COMPARISON OF AVERAGE YIELD OF OATS, WHEAT, AND BARLEY
HARVESTED FROM BORDER Rows AND CENTRAL Rows OF PLOTS 132 X 8,5 FT.
Oats
Wheat
Barley
Source
Num-
Yield
Num-
Yield
Num-
Yield
ber of
per acre,
ber of
per acre,
ber of
per acre,
plots
bu.
platH
bu.
plots
bu.
Outside border rows
44
132.0
20
55.0
16
97.7
Inside border rows
44
88.0
20
41.0
16
64 5
Central 13 rows . .
44
71.4
20
27.5
16
42.9
298 METHODS OF PLANT BREEDING
It is clear that border effect may profoundly influence yield.
Not all varieties were affected alike by surrounding alleys. The
rank order in yield frequently may not be the same with and with-
out border rows removed. Arny (1922) found that sowing winter
grains in the spring in the alleys between spring-grain plots
reduced border effect somewhat but did not prevent it completely.
Removing the outside two rows, 6 in. apart, before harvesting
for yield removes the larger part of the border effect and gives a
more accurate idea of the expected yields under good farm
conditions.
In variety trials in small plots, usually no alley is left between
plots. Varieties with different habit of growth consequently
grow adjacent to one another. Hayes and Arny (1917) demon-
strated that the border rows of tall varieties in three-row plots
replicated three times that grew adjacent to short varieties
yielded more than the central row. Often the intervarietal
competition was sufficient to cause differences of 4 or 5 bu. per
acre in yield of the border rows as compared with the central
row of the same variety.
Kiesselbach (1918) gave some illuminating information of the
effect of intervarietal competition in small grains. He compared
the differences in yield of adjacent single-row plots of different
varieties with the differencevS in yield of the same varieties grown
in alternate blocks, each consisting of three to five rows. The
yield of border rows was in some instances included in the yield
of the blocks. His results are summarized in Table 38.
The data illustrate clearly that competition between adjacent
varieties in one-row plots may seriously disturb the yields of
these varieties. In general, the varieties grown in alternate
rows show greater differences than when grown in alternate
blocks of multiple-row plots. The higher yielding varieties
usually benefit by planting in adjacent single rows, but this is
not always the case.
Intervarietal competition can be overcome by planting
multiple-row plots and discarding the bprder rows before harvest-
ing. Three- to five-row plots with one border discarded on each
side are commonly used. At Minnesota, the three-row plot
has become the standard for rod-row trials. Grouping the
material so that only strains of similar habit of growth and
maturity are grown adjacent to one another will tend to reduce
FIELD-PLOT TECHNIC
299
the effect of competition and permit harvest of the entire plot.
Such procedure implies, however, that the experimenter can
determine in advance whether competition will or will not be a
disturbing factor, an assumption for which there is no evidence,
usually, prior to the time the experiment is conducted.
TABLE 38. SUMMARY OF RELATIVE GRAIN YIELDS OF VARIETIES TESTED
IN SINGLE-ROW PLOTS AND ALSO IN BLOCKS CONTAINING SEVERAL Rows
Varieties compared in alternating
rows and in alternating blocks
Year
of
test
Ratio of variety 1 to variety 2
Alter-
nating
rows
Alter-
nating
blocks
Com-
peting in
same hill
Turkey Red (1) and Big Frame (2)
winter wheat
1913
1914
1913
1914
1913
1914
1913
1914
1912
1914
1914
1916
100:107
100:85
100:107
100:63
100:130
100:139
100:82
100:89
100:66
100:38
100:90
100:31
100:97
100:97
100:107
100:85
100:112
100:101
100:77
100:93
100:85
100:53
100:98
100:37
100:47
100:26
100:99
100:21
Turkey Red (1) and Big Frame (2)
winter wheat
Turkey Red (1) and Nebraska No. 28
(2) winter wheat
Turkey Red (1) and Nebraska No. 28
(2) winter wheat
Kherson (1) and Burt (2) oats
Kherson (1) and Burt (2) oats
Kherson (1) arid Swedish Select (2)
oats
Kherson (1) and Swedish Select (2)
(2) oats , ,
Hogue's (1) and Pride of the North
(2) corn . .
Hogue's (1) and Pride of the North
(2) corn
Hogue's (1) and University No. 3
(2) corn
Crossbred Hogue's (1) and inbred
Hogue's (2) corn
Tysdal and Kiesselbach (1939)" found that plots of alfalfa, of
the varieties Hardistan and Ladak, drilled in rows 7 in. apart,
were definitely subject to serious interplot varietal competition.
This could be overcome by removal of border rows. They found
little or no differential interplot competition in rows 12 in. apart.
Immer (1934) reported the results of tests with two varieties
of sugar beets, Old Type and Extreme Pioneer. These two
300
METHODS OF PLANT BREEDING
varieties were grown in alternate single-row plots and alternate
four-row plots, with the central two rows harvested. The rows
were 22 in. apart in 1930 and 1931 and 20 in. apart in 1932. As
an average of 10 replications in each of 3 years, Old Type yielded
3.78 0.44 tons more than Extreme Pioneer in single-row plots
but only 1.78 0.31 tons more in four-row plots. The difference
between these two differences was 2.00 0.54 tons. The higher
yielding variety profited at the expense of the lower in single-row
plots as compared with a multiple-row plot from which the border
rows were discarded.
The effect of competition within and between hills of corn is
frequently very striking. Kiesselbach (1923) has furnished some
pertinent information on the relative yields of one- , two- , and
three-plant hills adjacent to or surrounded by hills of variable
stand. These data are given in Table 39.
TABLE 39. RELATIVE YIELDS OF ONE-, Two-, AND THREE-PLANT HILLS
ADJACENT TO OB SURROUNDED BY HILLS OF VARIABLE STAND
Total
I\ T umber
Type of comparison
number
of hills
of ears
per 100
yield,
averaged
plants
per cen
3-plant hills surrounded by 3-plant hills . . .
598
89
100
2-plant hills surrounded by 3-plant hills . . .
120
99
82
1 -plant hills surrounded by 3-plant hills . .
80
141
61
3-plant hills adjacent to 1 hill with 2 plants
360
91
102
3-plant hills adjacent to 1 hill with 1 plant.
302
94
107
3-plant hills adjacent to 1 blank hilll
366
94
114
The four hills adjacent to a blank hill increased by 4 X 14 = 56
per cent. Therefore, only 44 per cent of the potential yield of
the missing hill was lost. This recovery is without consideration
of the slight increase that would be expected from hills on the
corners of the blank hill. A one-stalk hill surrounded by three-
stalk hills yielded but 61 per cent as much as three-stalk hills
surrounded by three-stalk hills, resulting in a loss in yield of
39 per cent. The four three-plant hills adjacent to a one-plant
hill increased in yield by 4 X 7 = 28 per cent of normal and
resulted in a net loss of 39 28 = 11 per cent because of the
inclusion of a one-stalk hill and the four adjacent three-stalk
FIELD-PLOT TECHNIC
301
hills in the yield determination. In the case of two-stalk hills,
the loss in yield was 18 per cent, but 8 per cent of this was
recovered in the four adjacent three-stalk hills.
Brewbaker and Immer, in Minnesota (1931), studied the effect
of missing hills, or hills with reduced stands, in inbred lines of
corn and F\ crosses. The data for an average of 2 years for the
FI crosses are given in Table 40. The average yield of the check
hills was 76.3 bu. per acre in 1928 and 78.4 bu. in 1929. The yield
of three-stalk hills surrounded on the four sides as well as four
corners was used as a check for the other yield comparisons.
TABLE 40. EFFECT OF COMPETITION WITHIN AND BETWEEN HILLS OF FI
('ROSSES OF CORN
Yield in
Number
Type of comparison
of hills
percentage
of check
3-phint hills between 2 blank hills
117
108 9
3-plant bills opposite 1 blank hill
78
105 6
3-plaut hills blank hills on 2 comers
69
100 2
3-plant hills between two 1-stalk hills
94
104 5
3-plant hills opposite one 1-stalk hill
72
103 7
3-plant hills between two 2-stalk hills
76
102 5
3-plant hills opposite one 2-stalk hill
77
101.5
2-plant hills surrounded by 3-plant hills
87
75 4
1-plant hills surrounded by 3-plant hills .
90
41
Each of four three-stalk hills adjacent to a blank hill increased
in yield by 5.6 per cent. The increase in yield of each of the four
adjacent three-stalk hills due to one- or two-stalk hills on one
side was 3.7 and 1.5 per cent, respectively. One-stalk hills and
two-stalk hills yielded but 41 .0 and 75.4 per cent, respectively,
as much as three-stalk hills. It follows, therefore, that the
inclusion of one- or two-stalk hills introduces *a greater error
than to ignore their effect on the yield of the surrounding three-
plant hills. Harvesting only hills with a perfect stand and
surrounded by perfect-stand hills sometimes reduces the number
of hills available for harvest materially, resulting in an increase
in the intraplot error. A part of the variation (frequently the
major portion) in stand between F\ crosses is due to random
causes and will tend to equalize over an average of the several
302 METHODS OF PLANT BREEDING
replications. In view of these considerations, a working rule has
been adopted at Minnesota of harvesting only three-plant hills
surrounded by one- , two- ; or three-plant hills in yield tests with
FI crosses. A further reason for this procedure is that frequently
one wishes to obtain the potential yield of the crosses without the
influence of differential stand resulting from differences dependent
upon the ability to produce a stand as influenced by different
inbred lines.
In yield tests with three-way or double crosses that are to be
grown commercially, a frequent practice is to determine the
percentage stand of each hybrid in the yield tests, but the yields
are based on the actual production of the plots without correction
for stand.
Kiesselbach and Weihing (1933) studied the effect on yield
of variable number of plants per hill in plots all having the same
total number of plants, under Nebraska conditions, to determine
the effect of variable stand on yield. The corn was planted with
an average of three plants per hill in the plots. The treatments
consisted of uniform three-plant hills and variable stands of
2-4, 1-3-5, and 1-2-3-4-5 plants per hill. The average yields of
these four treatments, for a 14-year period, was 49.9, 50.6, 49.3,
and 50.0 bu. per acre. It appears from these data that the yield
per plot is essentially the same provided that the total stand is
the same.
SIZE AND SHAPE OF PLOTS
In general, there are two kinds of experimental plots. Nursery
plots usually are small and are planted by hand or with special
nursery equipment and are cared for by hand cultivation. Field
plots generally are larger and are adapted to the use of standard
farm machinery. The distinction between these two types of
plots may, in some instances, be more or less arbitrary.
For cereal gra&ns, the rod-row plot is relatively standard in this
country. Such plots usually are 18 ft. long, and the plants to a
distance of 1 ft. at both ends of the plots are removed before
harvest. In the case of wheat, the yield in grams of a single row
16 ft. long with 1 ft. between rows multiplied by 0.1 converts the
yield to bushels per acre. Rod-row plots vary, in different types
of experiments and in different stations, from single to five-row
plots. In the case of multiple-row plots, one border row from
FIELD-PLOT TECHNIC 303
each side of the plot frequently is discarded to correct for possible
differential intervarietal competition. Rod-row plots frequently
are sown at the same rate of seeding recommended for com-
mercial farm planting in that area, although a lighter rate of
seeding is practiced by some workers. The distance between
rows usually is greater than that used by commercial farm drills,
since hand cultivation between the rows is needed to control
weeds satisfactorily.
Field plots vary from about ^{oo to J'lo acre in size. They
offer somewhat greater opportunity than single or three-row
plots to observe crop behavior under conditions comparable with
those found on farms. In general, experience has shown that
rod-row tests and large field plots compare very favorably in
testing for varietal differences, provided that adequate precau-
tions are taken to guard against competition and other errors.
Increasing plot size will, in general, decrease the error of a
single-plot yield. On the other hand, increasing plot size will
increase the land area in the blocks and, consequently, soil
heterogeneity in the blocks. The variability among sets of plots
of varying size will depend on the balance between these two
opposing tendencies.
Studies of the variability among plots varying in size and shape
are numerous. It is found generally that increasing replication
will decrease the standard error more rapidly than increasing
size of plots. Plots relatively small, adapted to the type of
nursery equipment available, should be used. The size used will
vary with the crop and conditions of the test. Field trials under
irrigation, for example, frequently require some modification in
procedure not necessary in tests without irrigation.
In general, it appears that long and narrow plots lead to a lower
error than square ones. Christidis (1931), from theoretical
considerations, suggested that long, narrow plots would control
soil heterogeneity better than plots more nearly square, occupying
the same area of land. He examined data from six uniformity
trials and found support for this hypothesis. Others have found
essentially the same thing. The relative efficiency of plots of
varying shape will depend on the direction of the fertility contour
lines across the field. If the predominant direction of these
contours is known, long narrow plots planted at right angles to
the direction of the fertility contours would lead to the low-
304 METHODS OF PLANT BREEDING
est error, since variability within blocks is then reduced to a
minimum.
In tests involving varieties of naturally cross-pollinated crops ?
the size of plot needed to obtain a given precision will depend
somewhat on the nature of the material. Bryan (1933) found
that about one-half as many plants or hills of corn were needed
to obtain the same level of precision in tests of the yielding ability
of crosses between inbred lines as compared with open-pollinated
varieties. Since the plants from crosses between inbred lines
are less variable than plants from open-pollinated varieties, the
variability within plots of hybrids will be reduced, and this, in
turn, will decrease the calculated variability between plots.
REPLICATION
Replication serves two purposes. (1) Replication increases
the precision of the experiment, since the mean of several replica-
tions provides a more accurate measure of varietal performance
than does a single plot. (2) From replicated trials, it is possible
to calculate an estimate of error of the experiment.
The number of replications used will depend on the variability
of the soil, the variability of the material to be tested, the degree
of precision desired, and the amount of seed available.
In experiments in which randomized blocks are used, the
standard error of the mean will be s/\/N, where 8 is the standard
error of a single determination and N is the number of plots of
each variety or treatment. The standard error of the mean may
be reduced, therefore, to whatever level is desired through
sufficient replication. After the standard error of a single
determination for plots of the desired size and shape is learned,
the number of replications needed to reduce the standard error
to a certain degree of accuracy may be obtained by the formula
/Standard error of a single determination^ 2
\ Standard error desired /
METHODS OF MAKING YIELD TRIALS USED IN MINNESOTA
If careful notes are taken on all characters such as lodging,
winter hardiness, and reaction to diseases and if varieties or
strains that are inferior to the standard variety in any important
FIELD-PLOT TEC UN 1C 305
respect are freely discarded before making extensive yield trials,
this, in many cases, will reduce the number of strains that must
be studied for yielding ability.
Where only a limited number of varieties or strains are included,
randomized blocks have considerable advantage over other
methods. The number of varieties included in each block should
be sufficiently small so that all varieties in each block are under
similar conditions. At Minnesota, 25 or fewer varieties in each
block have given relatively satisfactory results. One or two
standard varieties are grown in each block where more than a
single group of 25 varieties are under trial, and the standard
error is computed separately for each group of randomized
blocks. The methods of making yield trials may be illustrated
as they are carried on at Minnesota for spring wheat.
First Year. Rod-row trials are made at University Farm, St.
Paul, in single-row plots with two to four replications, depending
upon the amount of seed available. In these studies, check plots
of standard varieties are grown in each randomized block and in
each of the three disease nurseries: (1) for studies of reaction to
stem and leaf rust, (2) for studies of reaction to bunt, root rot,
and black chaff, and (3) for studies of reaction to scab. For the
scab trials, the rows are grown under a tent, which increases
humidity and makes infection relatively heavy. Notes are taken
on agronomic characters in the yield trials, and all varieties
or strains inferior to the check in any important respect are
eliminated.
Second Year. Rod-row trials are made at University Farm
in randomized blocks with not more than 25 varieties per block
and the inclusion of two or three standard varieties in each block.
Three-row plots replicated four times are used, the central row
of each plot being harvested for the yield trial and the border
rows for milling and baking tests. The data are analyzed by
means of the analysis of variance to determine if there are signifi-
cant differences in yield. A calculated standard error is obtained
by dividing the standard error of a single determination by the
square root of the number of replicates, i.e., randomized blocks.
Significant differences are considered to be twice the standard
error of a mean difference. In these and in subsequent trials,
all varieties are grown also in the disease nurseries and are
eliminated when found inferior to the standard varieties in any
306 METHODS OF PLANT BREEDING
important respect. The more promising varieties are tested also
in milling and baking trials.
If the number of strains available for testing is not too great
and there is sufficient seed, they may be tested the first or second
year at several stations.
Third to Fifth Year. Three-row plots are used as in the second
year, and trials are made in randomized blocks with three repli-
cations at each of four stations, University Farm, St. Paul, and
the branch stations at Morris, Crookston, and Waseca, making
12 replications in all. The data taken are similar to that out-
lined for the second year at University Farm. A mixture of seed
from the four stations is used for the milling and baking tests.
Sixth to Eighth Year. The more promising varieties in the
rod-row trials are increased and placed in )4(r acre plt trials in
randomized blocks, with three replications at each station, tests
being conducted at the same four stations and, in addition, at
Grand Rapids and Duluth. Very promising varieties in the
rod-row trials may be advanced to the ^o-acre plot trials before
the sixth year. Fewer than 20 varieties are in these yield tests, as
a rule, even though several thousand plant selections are grown
yearly. This means that varieties are discarded freely.
The standard errors for each trial are averaged for a series of
years and stations, considering that each trial may be a test of
the desirability of the new variety for use in some region in
Minnesota. The formula for the standard error of an average
that is used is 1/N -\/ sf + si s, where s\, si, etc., are errors
for each season and station and N is the number of trials made.
This generalized average error is used as a standard error for the
average yield of the varieties at several stations and for several
years.
Varieties grown in similar trials are then compared on the
basis of a standard error of a difference. The error is calculated
by multiplying the generalized average error by \/2. The differ-
ence in yield between any two varieties is obtained and the
significance of this difference determined by the t test, as
explained previously.
CHAPTER XX
RANDOMIZED BLOCKS, LATIN SQUARES, AND x 2 TESTS
Experimental designs and statistical methods that are in most
common use by the plant breeder will be illustrated for the
beginning student who has not had extensive mathematical
training or previous experience in the application of biometrical
methods to plant-breeding problems.
TESTS IN RANDOMIZED BLOCKS
One of the simplest experimental designs for testing the yield-
ing capacity of a group of varieties, particularly if the number is
not unduly large, say less than 25, is that of randomized blocks
In such designs, the varieties are grown in random order in each
of several complete replication series or blocks, the number of
replications used depending on the degree of precision desired
for the comparisons of the variety means.
R. A. Fisher stated that randomization of the order of varieties
in a block must be followed if an unbiased estimate of error is to
be obtained. Tedin (1931) tested the validity of this assumption
and confirmed Fisher's conclusion. Before describing studies in
randomized blocks, a method of obtaining a randomized order for
planting will be given where 20 varieties are being compared.
The following plan is one suggested by Fisher (1937) in "The
Design of Experiments." Use a pack of cards numbered from
1 to 100, and arrange them in random order by repeated shuffling.
If 20 varieties are to be tested, they are then numbered from 1 to
20 and cards drawn from the pack. Divide the number of the
card drawn by 20, and the remainder will give the variety to be
planted in the first plot. Suppose that the first card drawn is
33. Dividing by 20 leaves 13 as a remainder, and variety 13
is taken first. Suppose that the second card is 40, giving zero
as a remainder. Numbers divisible by 20 correspond to variety
20. Cards are drawn in this manner until the order of all varie-
ties has been obtained. The remainder corresponding to any
307
308
METHODS OF PLANT BREEDING
variety is disregarded after its first occurrence in the block.
After one block arrangement has been completed, the cards are
reshuffled before drawings are made for the second block.
Since 100 is divisible by 20, each variety will be represented in
the pack five times. If 19 varieties were to be placed in random-
ized order, the same pack of cards could be used but cards with
numbers above 95 discarded, since 95 is directly divisible by 19,
leaving no remainder.
Tables of random numbers, when available, such as those
given by Tippett (1927) and Fisher and Yates (1938) can be used
instead of a pack of cards in order to save labor. In such tables,
one may start at any point and proceed in any direction, taking
each pair of digits to represent the numbers of a card in a pack of
100. In these tables, 00 will be used in place of 100.
The analysis of variance is used to determine the significance
of results obtained in raiidomized-block designs. By this pro-
cedure, developed by Fisher (1938), the total variation is sepa-
rated into a number of components attributable to known or
controlled sources of variation and leaving a residual portion
due to uncontrolled causes and called the error.
Data from a randomiaed-block trial with 10 varieties of barley,
reported by Immer, Hayes, and Powers (1934), will be used to
illustrate the computation. The data are given in Table 41.
TABLE 41. YIELDS IN BUSHELS PER ACRE OF 10 VARIETIES OF BARLEY
GROWN AT UNIVERSITY FARM, ST. PAUL
Variety
Block number
Sum
Average
1
II
III
Manchuria
29 2
44 6
33 9
36 7
41 2
45.8
35 S
38 5
15.5
44.3
25
39 1
39 4
41
31 9
38.8
36.0
29 6
32.8
37.4
26 8
45 5
32.1
42
36 6
45 2
38
30 2
25 7
36.2
81
129.2
105 4
119 7
109 7
129.8
109 8
98 3
74.0
117.9
27.0
43.1
35.1
39 9
36 6
43.3
36 6
32.8
24 7
39.3
Glabron
Svansota
Velvet
Trebi
Minn. 457
Minn. 462
Peatland
Minn. 475
Barbless
Sum
365.5
351.0
3&8 3
1074.8 /
RANDOMIZED BLOCKS, LATIN SQUARES, AND x 2 TESTS 309
First it will be necessary to calculate the sum of squares of
deviations for the total variation and for blocks and varieties.
The sum of squares for error is obtained by subtracting the sum
of squares for blocks and varieties from the total.
The total sum of squares will be given by S(x 2 } S(x)x.
Squaring the 30 individual-plot yields and summing gives
S(x 2 ) = 39,949.06. The total, or S(x), = 1074.8. The mean
x = 1074.8 -T- 30 = 35.826,667. Then, the correction term
S(x)x will be 1074.8 X 35.826,667 = 38,506.50. Care must be
used to calculate the mean to sufficient accuracy. This difficulty
is overcome by squaring the total and dividing by the total
number of plots. Thus, ^^ = ^^ = 38,506.50. The
total sum of squares will be 8(x' 2 ) S(x)x = 39,949.06 -
38,506.50 = 1442.56.
The sum of squares for blocks is obtained by adding the squares
of the block totals, dividing by the number of plots in each
total, and subtracting the correction term. Expressed as a
S(x 2 )
formula, this will be TT?- ~ S(x)x, where x b refers to the block
totals. Numerically, this is 386,170.14 _ 38j506 50 = 10 51
The sum of squares for varieties is calculated in a manner
similar to that for blocks. If x v represents the variety totals,
118,668.36
^ * -111 , ._ ,. .
the sum of squares will be Q &(x)x. I his is -$ -
o o
- 38,506.50 = 1049.62.
In these and other calculations where totals are used, it is
necessary to calculate the sums of squares on a unit basis. This
is accomplished by dividing the squares of the totals by the
number of unit plots in each before subtracting the correction
term.
The complete analysis of variance is given in Table 42.
The degrees of freedom for blocks, varieties, and total are 1 less
than the number of blocks, varieties, and total plots, respectively.
The degrees of freedom for error will be the remainder after sub-
traction of the degrees of freedom for blocks and varieties from
the total. In randomized-block trials, the error degrees of free-
dom will be the product, also, of the degrees of freedom for
blocks multiplied by the degrees of freedom for varieties.
310
METHODS OF PLANT BREEDING
The column of mean squares in Table 42 is obtained by dividing
the sums of squares by the appropriate degree of freedom.
These values are the estimated variances expressed on a single-
plot basis. The standard error of a single plot is the square root
of the error mean square, i.e., -\X21.25 = 4.61.
TABLE 42. ANALYSIS OF VARIANCE OF YIELDS OF 10 VARIETIES OF BARLEY
IN A RANDOMIZED BLOCK TRIAL
Variation due to
Degrees of
freedom
Sum of
squares
Mean
square
F
s
Blocks
2
10.51
5.26
25
Varieties
9
1049 62
116 62
5 49*
Krror
18
382 43
21 25
4 61
Total
29
1442 56
* Exceeds the 1 per cent point.
To determine whether block or varietal differences are signifi-
cant, the variance ratios, or values of F, are calculated. These
are the mean squares for blocks and varieties each divided by the
error mean square. For comparing variety and error mean
squares, use is made of the table of F published by Snedecor
(1940) and reproduced as Appendix Table II. The expected
value of F for n\ = degrees of freedom for the larger mean
square and n^ = degrees of freedom for the smaller mean square,
or in this case for n\ = 9 and n 2 = 18, is found to be 2.46 and
3.60 at the 5 and 1 per cent points, respectively. The observed
value of F) for comparing variety arid error mean square,
exceeded the 1 per cent point. Therefore, we may conclude that
less than once in 100 trials could varietal differences as great as
those observed be obtained by chance. We may say, therefore,
that some of the varietal differences are highly significant, using
the term highly significant when the observed value of F exceeds
the 1 per cent point.
The mean square due to blocks was less than that for error.
This shows that the variation between block totals was less than
expected through random sampling alone. Whenever the mean
square for blocks or varieties is less than error mean square, there
is no point in calculating and testing the significance of F, The
removal of variability due to blocks removes differences due to
RANDOMIZED BLOCKS, LATIN SQUARES, AND x 2 TESTS 311
the placing of one block on land of different productivity from
another. This makes it possible to place each block on a sepa-
rate field if one so desires.
Standard malting barley varieties in Minnesota are Manchuria,
Velvet, and Barbless. The question may be asked, what are the
chances that these three varieties were significantly different in
yield in this test? The mean yields of Manchuria, Velvet, and
Barbless were 27.0, 39.9, and 39.3 bu. per acre, respectively. The
standard error of a mean of three plots is obtained by dividing
the standard error of a single determination (s) by \/N, where
N is the number of replications. The standard error of the
difference between two means of three plots is obtained by multi-
plying the standard error of the mean by \/2. This is on the
basis that the error in bushels, as calculated, will be the same for
all varieties regardless of their mean yield. The standard error
of the difference between two means may be calculated from
'2 X 21 25
- = 3.76 bu., where N is the number of plots
@_ /!
\ N V 3
in each mean. To determine the minimum difference required
for odds of 19.1 or 99 : 1, we must multiply the standard error of
the difference by the value of t for the degrees of freedom for error
in the analysis of variance at the 5 and 1 per cent points, respec-
tively. It is seen in Appendix Table I that for 18 degrees of
freedom t = 2.10 at the 5 per cent point and 2.88 at the 1 per cent
point. Multiplying 3.76 by 2.10 and 2.88 gives 7.90 and 10.83,
respectively. Yields of Velvet and Barbless were 39.9 and 39.3
bu. respectively, whereas Manchuria yielded only 27.0 bu. Since
Velvet and Barbless differed in yield from Manchuria by 12.9
and 12.3 bu., respectively, we may state that the chances are
less than 1 in 100 that these differences in yield are due to
chance.
If, instead of comparing the differences between these
varieties, irrespective of the direction of the difference, we want
to determine the probability that Velvet and Barbless exceeded
Manchuria in yield, we should need to use one tail only of the
probability curve. In that case, we should use t for P = .10
and .02 instead of .05 and .01 to obtain the probability of a devi-
ation in one direction only occurring in 5 and 1 per cent of the
time through random sampling.
312 METHODS OF PLANT BREEDING
The plant breeder usually compares the performance of new
strains with standard varieties whose performance has been well
established. If the odds were 19.1, for example, that the differ-
ence in yield between Barblcss and Manchuria, plus or minus,
are not due to chance, the odds would be 39 : 1 [(2 X 19 + 1) : 1]
that Barbless exceeded Manchuria. The odds that the increased
yield of Barbless over Manchuria is not due to chance would
be two times, plus one, that found from the t table.
Some practices essential to satisfactory randomized-block tests
can be summarized at this point. Size of plot used may be
decided on two general bases. Sufficient plants to give a measure
of the characters of the strain or variety and a plot of convenient
size from the standpoint of handling and cost are the bases for
selection of a plot of proper size. In general, the number of
varieties in a block should not be much greater than 25, or the
area of the block will be so great that soil heterogeneity may
lead to too large a standard error. In simple plant-breeding
experiments, when large numbers of strains are available for
testing, the trials may be made by using randomized blocks with
not more than 25 varieties in a block and the growing of one or
more standard varieties in each block., New strains are com-
pared with the standard by this method before comparing those
that survive this test with each other.
LATIN SQUARES
The Latin square has been shown to be a desirable method of
making precise comparisons when the number of treatments
or varieties to be compared is small, i.e., from about 4 to 10.
Although of less general value for the plant breeder than random-
ized blocks, it is a desirable method for special experiments.
In a Latin-square design, there are as many replications as
treatments. The treatments are arranged in a random order
in a square or rectangle, with the restriction that each treatment
can occur but once in each row and each column. For randomiz-
ing the order of the varieties, use may be made of a pack of cards,
as described for randomized blocks, or by reference to the pub-
lished sets of Latin squares [from 4 X 4 to 9 X 9, given by Fisher
and Yates (1938)]. A Latin-square arrangement for five treat-
ments, A, J? ? C ? D ? E, is illustrated below;
RANDOMIZED BLOCKS, LATIN SQUARES, AND % 2 TESTS 313
E B D C A
A C E B D
B E A DC
C D B A E
D A C E B
The degrees of freedom for an analysis of variance would be
keyed out as follows:
Degrees of
Variation Due to Freedom
Rows 4
Columns 4
Treatments 4
Error 12
Total 24
The calculation of the sums of squares proceeds as outlined
for randomized blocks. If x = the yield of each plot, x r) x c , and
Xt the total yield of each row, column, or treatment, and p = the
number of rows, columns, and treatments, the sum of squares
for rows, columns, and treatments and the sum of squares for the
total may be calculated as follows :
Sum of squares for rows - S(x)x
P
Sum of squares for columns _JL _ fi( x )
P
Sum of squares for treatments 8(x)$
P
Sum of squares for total S(x 2 ) - S(x)x
The sum of squares for error is obtained by subtracting the sum
of squares for rows, columns, and treatments from the total.
The shape of plots in a Latin square need not be square. They
may be rectangular in shape. If very long and narrow, however,
the variation in soil fertility in the narrow direction will be small,
and little will be gained by removing the variation in this direc-
tion. With plots for which the ratio of length to width is not
extreme, removing variability in two directions usually will
result in reduction in the error as compared with randomized
blocks,
314 METHODS OF PLANT BREEDING
ESTIMATING THE YIELD OF A MISSING PLOT
It sometimes happens that the yield of some plot, or plots, is
lost or is known to be unreliable. When that happens, it may be
desirable to interpolate yields for the missing plots before the
analysis of variance can be completed.
Yates (1933) has given a formula for estimating the yield of a
missing plot. The interpolated yield is so calculated that the
error variance is made a minimum. The formula appropriate
for randomized-block trials is as follows:
pP + qQ-T
(p- l)(q - 1)
where X = yield of missing plot.
p = number of treatments.
q = number of blocks.
P sum of known yields of treatment with a missing plot.
Q = sum of known yields of block with a missing plot.
T = total yield of known plots.
As an example, assume that the yield of Minn. 462 in block II
of Table 41 were missing. Then
p = 10
g = 3
P = 73.8
Q = 315.0
T = 1038.8
v - (10 X 73.8) + (3 X 315.0) - 1038.8 _
-- -
which is the estimated yield of the missing plot.
If two or more plots are missing, a method of approximation
may be used based on the foregoing formula. With, say, two
plots missing, a yield is assumed for one of the missing plots and
the formula used to estimate the second. The assumed yield
of the first missing plot is then erased and the formula used to
estimate that. The yield estimated first is then recalculated and
the same procedure applied in rotation to obtain the accuracy
desired.
After obtaining the estimated yield of the missing plot, the
analysis of variance is carried through in the usual manner except
that 1 degree of freedom is subtracted from the error and 1 from
RANDOMIZED BLOCKS, LATIN SQUARES, AND x 2 TESTS 315
the total for each missing plot interpolated. In plant-breeding
trials in randomized blocks, the data for the variety in which one
or more plots are missing may be disregarded in computing the
analysis of variance, providing that the degrees of freedom used
are for the varieties actually used in the analysis.
In Latin squares, the formula for estimating the yield of a
missing plot is
P (P r + P c + P t ) ~ 2T
(p- l)(p - 2)
where X = missing plot yield.
p = number of rows, columns, or treatments.
P r , P c , P t = sum of the known yields of the row, column, or
treatment with a missing plot.
T = total yield of known plots.
SPLIT-PLOT EXPERIMENTS
In designing experiments involving two or more factors, the
split-plot arrangement is often useful. The arrangement of the
plots and the mariner of calculation of the data will be illustrated,
with the use of data obtained by A. C. Arny.
This experiment was one designed to determine the effect of
varying the width between rows and spacing of seed within rows
on the yield of soybeans. Four-row plots 132 ft. long were
planted in rows 16, 20, 24, 28, 32, or 40 in. between rows. These
long plots were then divided into four subplots of 33 ft. each and
the soybeans spaced J^, 1, 2, or 3 in. apart within the subplots.
The entire experiment was replicated four times, and only the
central two rows of each four row plots were harvested. The
field arrangement of the plots in block III is given below:
16
1
3
2
32
1
3
2
M
28
3
1 A
1
2
40
2
1
M
3
24
1
3
2
M
20
3
1
2
316
METHODS OF PLANT BREEDING
The order of the plot widths within each block was random,
and the four spacings were randomized within each main or
width plot. The dimensions of each block were 53}i by
132 ft.
In Table 43 are given the yields in bushels per acre for each
plot, arranged in a convenient form for computation.
TABLE 43. YIELD OP SOYBEANS IN BUSHELS PER ACKE
Block
Width
of rows,
Spacing within rows, in.
Block
total
in.
M
1
2
3
Sum
I
16"
25 1
21.3
22.3
22.1
90 8
20"
21.8
22.7
22.2
22.8
89.5
24"
21.9
21.8
21 2
20.6
85.5
28"
21.2
20.4
20.4
17.9
79.9
32"
20.7
20.0
18.3
20.0
79
40"
19.5
18.3
17.5
16.3
71 6
496.3
11
16"
25.2
19.9
22.1
22 7
89 9
20"
21.9
21 3
22.1
22 9
88.2
24"
19 7
19 8
20 1
19.8
79 4
28"
20 8
21 2
18.8
20.6
81 4
32"
18.5
20.7
17 5
16.4
73 1
40"
18.5
18.2
19 8
15 9
72.4
484.4
111
16"
15 7
21.6
22.9
20.3
80 5
20"
22
20 4
22 4
20.7
85.5
24"
25 5
20.7
20 7
20 5
87.4
28"
21.5
19.9
20 5
20.9
82.8
32"
22.0
19.3
18.1
17.8
77.2
40"
20.5
16.4
17.5
18 5
72.9
486.3
IV
16"
23.8
29.0
12.3
23 5
88 6
20"
27.0
21.2
20.5
20.7
89 4
24"
23.5
20
22.3
19.8
85.6
28"
22 5
21.5
22 7
18 9
85.6
32"
23.9
18.4
20 7
18.7
81.7
40"
19 9
17 8
16.9
18.5
73.1
504
Sum
522.6
491.8
479 8
476.8
1971
1971.0
Squaring the 9G individual plot yields gives S(x 2 ) = 41,045.92.
The correction term will be S(x)x = (1971.0) 2 /96 = 40,467.09.
The total sum of squares is then 41,045.92 - 40,467.09 = 578.83.
RANDOMIZED BLOCKS, LATIN SQUARES, AND % 2 TESTS 317
The block sum of squares will be
S(xl) c , , 971,460.74
' S(x)x = - ^ -
1AAA
= 10.44
where x$ represents the squares of the block totals.
The sum of squares for the main or width plots is calculated
from the column of sums on the right-hand side of Table 43.
Thus,
90.8 2 + 89.5 2 + +73.1 2 a/ ..
162,768.54
- 40,467.09 = 225.05
The data in Table 43 are next assembled in the form given in
Table 44 in order to obtain the sums of the four replications for
each width of row and spacing within the rows.
TABLE 44. TOTAL YIELD FOR EACH WIDTH OF Row AND SPACING FOK
THE FOUR REPLICATIONS
Width of
Spacing within rows, in.
Sum
Average
rows, m.
>'2
1
2
3
16
89 8
91.8
79.6
88.6
349.8
21 9
20
92.7
85.6
87 2
87 1
352 6
22.0
24
90.6
82 3
84 3
80 7
337 9
21 1
28
86.0
83
82.4
78 3
329 7
20 6
32
85 1
78 4
74 6
72 9
311
19 4
40
78 4
70.7
71 7
69 2
290
18 1
Sum . . .
522 6
491 8
479 8
476 8
1971
Average. . .
21.8
20.5
20.0
19 9
The sum of squares for widths of row will be ~~ S(x)x =
650,386.30
16
16
- 40,467.09 = 182.05, where xl is the square of the
totals of 16 plots for each width.
The sum of squares for spacings will be ^-^ S(x)x =
972,524.28
24
- 40,467.09 = 54.75, where xl is the square of the
totals of 24 plots for each spacing.
318
METHODS OF PLANT BREEDING
Next, the 24 sums within Table 44 are squared and added.
Thus
89.8 2 + 92.7 2
+ 69.2 2
- 40,467.09 = 267.23 for
these 23 degrees of freedom. From this sum of squares is sub-
tracted the sums of squares for width and spacing to give 267.23
182.05 54.75 = 30.43 as the sum of squares for the interaction
of width X spacing.
The entire analysis of variance is given in Table 45.
TABLE 45. ANALYSIS OF VARIANCE OF YIELD IN BUSHELS PER ACRE IN-
SPACING TRIAL WITH SOYBEANS
Variation due to
Degrees of
freedom
Sum of
squares
Mean
square
F
Blocks
3
10 44
3.48
1.60
Width, of row
5
182 05
36 41
16 78 f
Error a
15
32.56
2 17 = s z a
Main plots
23
225.05
Spacing
3
54 75
18 25
3 67*
Width X spacing
15
30,43
2.03
Error b
54
268 60
4 97 = *J
Total
95
578 . 83
* Exceeds the 5 per cent point,
f Exceeds the 1 per cent point.
The degrees of freedom and sum of squares for error a are
obtained by subtracting the degrees of freedom and sum of
squares for blocks and widths from the main plots, which have
23 degrees of freedom and the sum of squares for which was
225.05. The degrees of freedom for error 6 will be 95 15
3 23 = 54. The sum of squares for error b is obtained by
subtraction in like manner.
It is noted that the value of F for a comparison of the mean
squares for widths of row with that of error a was highly signifi-
cant. The mean square for spacings is compared with that of
error b and exceeded the 5 per cent point but did not reach the
1 per cent point. The mean square for interaction of width X
spacing was numerically less than the mean square for error b and
clearly not significant.
RANDOMIZED BLOCKS, LATIN SQUARES, AND x 2 TESTS 319
From these data one may conclude that the differences in yield
of soybeans in this test, planted in rows of different width,
was independent of the spacing within the rows. The planting
arrangement that would be expected to result in the highest
yield would be the combination of highest average width per row
and spacing within the row.
The standard error of the difference between two means for
different widths of row would be
2~X"3_ /
~~16~ \
2X~2l7
~
16
si being divided by 16, since the means arc based on that number
of plots and multiplied by 2, since the standard error of a differ-
ence is desired. Since t at the 5 per cent point for 15 degrees
of freedom is 2.13, the minimum level of significance would be
2.13 X 0.521 = 1.11 bu. The average yield of all plots with
20-in. rows was 22.0 bu., and the average yield for 28-in. rows
was 19.4 bu. The difference of 1.4 bu. exceeded the minimum
level of significance and may be judged significant.
The standard error of the difference between the means of two
spacings would be
^ '~~ = 0.644
Since t at the 5 per cent point and 54 degrees of freedom is 2.00,
mean differences in excess of 2.00 X 0.644 = 1.30 may be judged
significant. Spacing of the soybeans 2 in. apart in the row
resulted in a reduction of 1.8 bu. per acre over J^-in. spacing, a
significant decrease.
Split-plot experiments are very useful when two or more factors
are to be tested in one experiment and planting difficulties make
it necessary that large plots be used for one factor, the large, or
main, plots being split up for the second factor. Variety tests
combined with dates of planting tests would be of this type, the
dates of planting being the main plots and the varieties the sub-
plots. Tests of f ungicidal dusts on different varieties would often
require rather large plots for the different dust treatments in
order to control "drift" of the dust in application. These could
well be the main plots and several varieties planted in subplots
320
METHODS OF PLANT BREEDING
within each main or dust treatment plot. For such tests, the
split-plot design would be well suited.
CHI-SQUARE (x 2 ) TESTS
Tests of Goodness of Fit. The x 2 test is a useful method for
testing goodness of fit of Mendelian ratios, as pointed out by
Harris (1912). *The methods will be illustrated by various
examples.
In the 7^2 generation of a cross between a two-rowed variety of
barley producing green seedlings (VV LgLg) with a six-rowed
variety with light-green seedlings (vv Iglg) the following number
of plants was obtained in the four phenotypic classes:
VLg
Vlg
vLg
v Ig
Total
281
59
60
58
458
The general formula for calculating x 2 niay be written as
2 s(o -
v * ~~ '
x c
where S refers to summation, is the observed frequency, and
C is the expected or calculated frequency.
The test for deviations of the single factor ratio Vv may be
made as follows :
Phenotype
Observed (0)
Calculated (C)
n r
(O - C) 2
c
V
v
Total
340
118
343.5
114 5
-3.5
3 5
0.036
0.107
458
458.0
x 2 = 0.143
On entering the table of % 2 (Appendix Table III) for 1 degree
of freedom, it is found that the observed x 2 lies between P = 0.95
and 0.50. The degrees of freedom are 1 less than the number of
classes. A x 2 value as great as the observed would be expected
to occur between 50 and 95 times in 100 trials through errors of
random sampling.
RANDOMIZED BLOCKS, LATIN SQUARES, AND % 2 TESTS 321
X 2 for a 3:1 ratio may be calculated also from
where A = observed number in the dominant class.
a = number in the recessive class.
N = total number.
For the foregoing problem
9 [340 - 3(118)] 2 A1 . ,, , ,, ,, ,
X 2 = - .L = 0.143, the same as found by the longer
o X 4oo
method.
Below are listed some formulas that may be useful for two
class segregations.
Segregation Expected x 2 Formula
A:a
1:1
2:1
3:1
15:1
9:7
The agreement between the ratio observed and the ratio
expected on the basis of independent inheritance of the two factor
pairs may be tested by calculating x 2 for goodness of fit to a
9:3:3:1 ratio. The calculations are given in Table 46, with the
use of the data from the experiment mentioned previously.
TABLE 46. CALCULATION OF GOODNESS OF FIT TO A 9:3:3:1 RATIO
x 2 = (A - a) 2 /#
x 2 = (A - 2a)*/2N
X 2 (A - 30
(A - 15a) 2 /15JV
(7A - 9
Phenotype
Observed (0)
Calculated (C)
- C
(0 - C) 2
c
V Lg
a = 281
257.625
23.375
2 121
Vlg
6 = 59
85.875
-26.875
8.411
vLg
c = 60
85.875
-25 875
7 796
v Ig
d = 58
28 625
29 375
30 145
Total
N = 458
458.000
000
x 2 - 48 473
The calculated frequency for the four phenotypic groups will
be %e? Mf>? Me? an d 3/fe of the total, respectively. On entering
the table of x 2 (Appendix Table III) for 3 degrees of freedom, it
is noted that the observed value of %! == 48.473 greatly exceeds
the 1 per cent point. In goodness-of-fit tests, such as the fore-
322 METHODS OF PLANT BREEDING
going, the degrees of freedom are 1 less than the number of
classes. It may be concluded, therefore, that the deviation
from a 9:3:3:1 ratio was highly significant.
A somewhat shorter method may be used for testing goodness
of fit to a 9 : 3 : 3 : 1 ratio. Thus
. 16(a 2 + 36 2 + 3c 2 Ar
X 2 = , - _ -- N
where a, 6, c, and d = observed frequencies as given in Table 46.
Ther
2 - 16[28P + 3(59 2 ) + 3(60 2 ) + 9(58 2 )] _
x -- 9 X458 4M ~~ ** A76
as before.
The nature of the deviation of the observed ratio from that
expected on the basis of independent inheritance may be deter-
mined oy separating x 2 into its components. The 3 degrees of
freedom for the goodness-of-fit test may be apportioned to: one
for deviat ons of the Vv segregation from a 3:1 ratio, one for
deviations of the Lg Ig segregation from a 3:1 ratio, and one
for detecting association (linkage) of the two factor pairs. Con-
venient formulas are
For Vv segregation x 2 = (a + 6 - 3c - 3d) 2 /3N
For Lg Ig segregation x 2 = (o - 36 + c - 3d)*/3N
For linkage x 2 = (a - 36 - 3c + 9d) 2 /9N
The formula for deviations of single-factor ratios will reduce to
the form given before. Substituting the observed ratios in these
three formulas gives
For Vv segregation x 2 = 0.143 for 1 degree of freedom
For Lg Ig segregation x 2 = 0.073 for 1 degree of freedom
For linkage x 2 = 48.257 for 1 degree of freedom
For goodness of fit x 2 =: 48,473 for 3 degrees of freedom
In referring to the table of x 2 (Appendix Table III), it is noted
that the agreement of the two single-factor ratios with a 3:1
ratio is good. The x 2 for linkage exceeds the 1 per cent point.
With the use of the product method for calculating linkage
(Fisher 1938) and tables provided by Immer (1930), the per-
RANDOMIZED BLOCKS, LATIN SQUARES, AND x 2 TESTS 323
centage of recombination between these two factor pairs was
30,2 2.7.
x 2 for Independence. The % 2 test may be used to determine
whether two characters, classified into two or more groups, are
independent. The calculations will be illustrated with data
obtained by Hayes, Moore, and Stakman (1939) in a study of
inheritance of characters in crosses between varieties of oats.
Among the characters studied were plumpness of grain and
type of awn. One of the parents, Bond, produced short, weak
awns; the other parent, designated Double Cross, produced long,
heavy awns. The two parents differed in plumpness of grain,
as may be noted in the following table. Plumpness of grain is a
visual note taken on a scale of to 100 and has been found to be
correlated significantly with yield.
TABLE 47. PERCENTAGE OF PLUMPNESS OF KERNEL OF INDIVIDUAL PLANTS
OF THE PARENT VARIETIES
Variety
Number of plants in plumpness classes
0-25
26-50
51-75
76-100
Total
Bond
1
5
6
28
54
26
61
122
59
Double Cross .
In Table 48 are given the number of F 2 plants in different
classes of plumpness and awn development.
TABLE 48. FREQUENCY IN CLASSES FOR AWN DEVELOPMENT AND PERCENT
OF PLUMPNESS OF GRAIN OF F 2 PLANTS IN THE CROSS OF
BOND X DOUBLE CROSS
Plumpness, per cent
Awn classes
Total
Weak
Intermediate
0-50
51-75
76-100
46
165
120
8
44
27
54
209
147
Total
331
79
410
Proportion
0.80732
. 19268
To determine whether these two pharacters are independent,
we may compare the observed frequencies with theoretical fre-
324
METHODS OF PLANT BREEDING
quencies calculated on the assumption of independence. The
theoretical frequencies for the individual cells of the table are
calculated so that they are in the same proportion to one another
as they are in the totals of the rows and columns.
The theoretical frequency in the upper left-hand cell of the
table will be the product of the two marginal totals divided by
the grand total or (331 X 54)/410 = 43.60. The other theoretical
frequencies are calculated in a similar manner. To expedite the
computations, the proportion of the grand total in each column
may be calculated first. This is designated as " proportion" in
Table 48. The proportion of weak and intermediate awn plants
may then be multiplied by the margin totals for plumpness
classes, i.e., 54, 209, and 147, to obtain the theoretical frequencies.
In Table 49 is given the computation of % 2 for independence.
TABLE 49. CALCULATION OF % 2 F R INDEPENDENCE OF AWN TYPE AND
PLUMPNESS OF KERNELS
Observed
frequency
Calculated
frequency
C
(0 - C) 2
C
46
43.60
2.40
.132
165
168 73
-3 73
.082
120
118 68
1 32
.015
8
10.40
-2.40
.554
44
40 27
3 73
.345
27
28.32
-1.32
.062
Sum 410
410.00
0.00
x 2 = 1 190
Since % 2 was calculated from fixed marginal totals, the degrees
of freedom are (r l)(c 1) = 2, where r and c refer to the
number of rows and columns, respectively, in Table 48. On
entering the table of x 2 for 2 degrees of freedom, it is found that
the observed x 2 gives a value of P between .50 and .70. These
data indicate, therefore, that there was no association between
plumpness of kernel and development of awn in this segregating
population. The x 2 test is frequently useful in testing for inde-
pendence of different characters in plant-breeding studies, when
the data for each character are grouped in classes and entered as
illustrated in Table 48.
CHAPTER XXI
CORRELATION AND REGRESSION IN RELATION TO
PLANT BREEDING
SIMPLE CORRELATION
When data on two or more characters of a group of varieties or
treatments are available, it frequently will be of value to deter-
mine the degree of association between them. This may be done
by calculating the correlation coefficient. The coefficient of
correlation can vary from +1 to 1, being zero when there is no
association and increasing to + 1 or 1 for complete association.
A method of computation will be illustrated by using data
obtained in Minnesota from a study of the relationship between
number of kernels per spikelet and yield of grain in bushels per
acre in rod-row trials with spring wheat. Each plot consisted of
three rows, and the central row only was used for obtaining the
number of kernels per spikelet and yield. The number of kernels
per spikelet was obtained from 100 spikes, selected at random,
from each plot. The yield was computed from the central row
of each three-row plot. The experiment was a randomized-block
trial. The data are given in Table 50.
Before making a study of the extent of correlation between
yield and number of kernels per spikelet, it will be of value to
determine whether these strains of wheat differed significantly in
the two characters being studied. This is accomplished through
an analysis of variance. The results are given in Table 51.
The mean squares for varieties compared with error exceeded
the 1 per cent point, indicating that highly significant differences
in yield and number of kernels per spikelet existed among these
21 varieties. For neither character were there significant differ-
ences between blocks.
To calculate the correlation coefficient, it is necessary to deter-
mine the co variance. This is computed from the sums of products
in a manner analogous to the computation of sums of squares. The
sums of products are obtained from the sum of the products
325
326
METHODS OF PLANT BREEDING
TABLE 50. NUMBER OF KERNELS PER SPIKELET AND YIELD IN BUSHELS
PER ACRE, IN EACH OF THREE REPLICATIONS, IN ROD-ROW TRIALS
OF SPRING WHEAT
Kernels per spikelet
Yield, bu.
1
2
3
Total
Mean
1
2
3
Total
Mean
Marquis
1 5
1 5
1 4
4 4
1 47
22 7
22 3
28 8
73 8
24 6
Ceres 4 . . ....
1 9
1
1 7
5 2
1 73
32 9
31 1
27 4
91 4
30 5
Hope
1 1
1 9
1 1
3 4
1 13
27 1
21 3
18 1
66 5
22 2
Ceres X Hope No. 1 ... ...
1 4
1 7
1 4
4 5
1 50
19 4
19 2
23 7
62 3
20 8
Ceres X Hope No 2
1 5
1 4
] 4
4 3
1 43
26 4
35 1
28 1
89 6
29 9
Ceres X Hope No. 3
1.5
1.4
1.3
4.2
1.40
26.2
36.4
29.3
91.9
30.6
Ceres X Hope No. 4
1.4
1.5
1.4
4.3
1.43
24.2
24.1
20.0
68.3
22.8
Ceres X Hope No. 5
1 4
1 4
1 3
4 1
1 37
23 g
26 3
24 3
74 4
24 8
Double Cross No. 80
1 3
1 ' ?
1 4
3 9
1 30
26 8
25 7
30 5
83
27 7
Double Cross No. 85
1 5
1 3
1 4
4 2
1 40
22 6
26 6
19 9
69 1
23
Double Cross No. 86
1.4
1.5
1.3
4.2
1.40
23.2
23.5
28.8
75.5
25.2
Double Cross No. 97
1 4
] 4
1 4
4 2
1 40
32 4
27 5
28 1
88
29 3
Double Cross No. 98
1 4
1 3
1 5
4 9
1 40
26 1
25 3
30 7
82 1
27 4
Double Cross No. 99
1.4
1.5
1,3
4.2
1.40
22.1
28.1
28.6
78.8
26.3
Double Cross No. 100
1.5
1.5
1.3
4.3
1.43
27.1
28.3
26.8
82.2
27.4
Double Cross No. 3 02
1.4
1.3
1.4
4.1
.37
27.1
30.8
28.9
86.8
28.9
Double Cross No. 103. ...
1.4
1.5
1.4
4.3
.43
26.9
29.1
22.6
78.6
26.2
Marquis X H44, No. 25
1.2
1.3
1.2
3.7
.23
15.9
18.7
19.8
54.4
18.1
Marquis X H44, No. 33
1 1
1.2
1 3
3.6
.20
27.9
27.9
23.5
79.3
26.4
Marquis X H44, No. 35
1.2
1.4
1.3
3.9
.30
27.0
21.4
25.0
73.4
24.5
Marquis X H44, No. 40
1.3
1.2
1.2
3.7
23
22.6
23.3
24.0
69.9
23.3
Total
29.2
29.3
28.4
86.9
530.4
552.0
536.9
1619.3
of the deviations of x and y from their means, which may be
expressed as S(x x)(y g). This is most conveniently cal-
culated from S(xy) S(x)S(y)/N, and the co variance is found
by dividing the sum of products by the appropriate degrees of
freedom. The sums of products may be either positive or
negative.
An analysis of covariance is made in the same manner as an
analysis of variance. Letting y and x represent yield in bushels
per acre and number of kernels per spikelet, respectively, the
total sum of products is given by S(xy) S(x)S(y)/N. Mul-
tiplying each plot yield by the number of kernels per spikelet for
that plot and summing gives 2242.16. Since S(x) = 86.9
and S(y) = 1619.3, the correction term will be obtained by
multiplying 86.9 by 1619.3 and dividing by 63, which gives
2233.606. The total sum of products will be 2242.16 - 2233.606
- 8.554.
CORRELATION AND REGRESSION 327
Letting x& and yb be the block totals for x and y, the sum of
products for blocks will be i^M^ - S(x)S(y)/N, or 2233.773
- 2233.606 = 0.167.
If x v and y v are the variety totals for x and y, the sum of
products for varieties will be ^ S(x)S(y')/N t or 2243.937
- 2233.606 = 10.331.
TABLE 51. ANALYSIS OP VARIANCE OF YIELD IN BUSHELS PER ACRE AND
NUMBER OF KERNELS PER SPIKELET OF 21 VARIETIES OF SPRING WHEAT
IN A RANDOMIZED-BLOCK TRIAL
Variation due to
Degrees of
freedom
Sum of
squares
Mean
square
F
Yield per acre (y)
Blocks
2
20
40
11 69
654.29
378 04
5 . 845
32.714
9.451
3.46*
Varieties . ...
Error
Total
62
1,044.02
Kernels per spikelet (x)
Blocks
2
20
40
023
0.930
330
0.0115
0.0465
0082
1.40
5.67*
Varieties
Krror
Total
62
1.283
* Exceeds the 1 per cent point.
The error sum of products is obtained by subtracting sums of
products for blocks and varieties from the total.
In the following table are assembled the sums of squares (from
Table 51) and sums of products for the different components of
the total variation together with the correlation coefficients.
The coefficient of correlation will be given by
sum of products of xy
\Xsum of squares of y \/sum of squares of x
The correlation between varieties is found to be
10.331
V654.29 V0.930
= +.419
328
METHODS OF PLANT BREEDING
The significance of a correlation coefficient may be determined
by reference to Appendix Table V, where the degrees of freedom
will be 2 less than the number of pairs. The variety correlation
+ .419 was slightly less than the value of r = .433 at the 5 per
cent point for 19 degrees of freedom.
TABLE 52. SUMS OF SQUARES AND PRODUCTS FOR YIELD IN BUSHELS PER
ACRE AND KERNELS PER S PIKELET AND THE CORRELATION COEFFICIENTS
Variation due to
Degrees
of
freedom
Sums of squares or products
r
y*
xy
X 2
Blocks ,
2
20
40
11.69
654,29
378 04
0.167
10.331
-1.944
0.023
0.930
0.330
+ .419
-.174
Varieties
Error .... .
Total
62
1,044.02
8.554
1.283
If it is desired to calculate a correlation coefficient for, say,
yield and kernels per spikelet of varieties without performing an
analysis of variance and co variance, a convenient formula is
r =
8(xy) - S(x)S(y)/N
- [S(x)]*/N
The calculations, with the use of the mean kernels per spikelet
and yield in bushels for the varieties in Table 50, will be illus-
trated. Multiplying the mean kernels per spikelet by the mean
yield for each variety and summing gives S(xy) = 747.773.
Summing the squares of the means for kernels per spikelet and
yield in bushels gives S(x*) = 40.2199 and S(y 2 ) = 14,098.53.
Since the total of the means are S(x) = 28.95 and S(y) = 539.9,
the correlation coefficient will be
r =
747.773 - (28.95) (539.9)/21
V 40.2199 - (28.95) 2 /21 V 14, 098.53
(539.9) 2 /21
= +.423
This agrees closely with r = +.419 obtained in Table 52, the
discrepancy being due to rounding off figures in recording the
means.
If this formula for the correlation coefficient used above is
multiplied by N/N
r =
CORRELATION AND REGRESSION 329
N8(xy\- S
which probably is the best form for rapid machine computation.
Numerically, this will be
21(747.773) - (28.95) (539.9) __
V2f(402199) - (28.95) 1 V^T^O^^- "(539:9) 2
= +.423
LINEAR REGRESSION
The relationship between two variables may be expressed also
by means of the regression coefficient. The regression coeffir
cient gives the rate of change in one variable (the dependent
variable) per unit rate of change in another (the independent
variable). The regression coefficient is given by
v * S(x - x) 2 S(x*) - [8(x)]*/N
This may be expressed also as
, _ sum of products of xy
yx sum of squares of x
where b yx = regression of y on x.
Numerically, the regression of yield on number of kernels per
spikelet for the 21 varieties in Table 52 is 10.331 -4- 0.930 =
+ 11.109. This means that as the number of kernels per spikelet
of the varieties increased by 1.0, the yield of the varieties, on the
average, increased by 11.1 bushels, or with an increase of spikelet
number of 0.1 yield in bushels increased by 1.11,
The significance of a regression coefficient may be tested by
means of an analysis of variance. The total sum of squares
for varieties is given in Table 52. The test of significance of
regression is given in Table 53. The sum of squares due to
regression for varieties will be
(Sum of products of xyY = (10.331) 2 = 76
Sum of squares of x 0.930
The value of F obtained, 4.04, fails to reach the 5 per cent
point (4.38) for ni = 1 and n<t = 19 degrees of freedom.
330
METHODS OF PLANT BREEDING
It is noted that in testing the significance of both the correla-
tion coefficient r and the regression coefficient 6, both were non-
significant. The tests for significance of r and b are equivalent.
When one is significant the other is significant also and vice
versa. Exactly the same probabilities are obtained by the two
tests Of significance.
TABLE 53. TESTING SIGNIFICANCE OF A REGRESSION COEFFICIENT
Variation due to
Degrees
of freedom
Sum of
squares
Mean
square
F
Regression
1
114.76
114 76
4.04
Deviations from regression . . .
19
539.53
28.40
Total
20
654 . 29
If the sum of squares due to regression is divided by the total
sum of squares, we may express this as a percentage of the total
sum of squares accounted for by regression. Such a quantity is
r 2 . The square of the correlation coefficient may be used as a
measure of the percentage of the total variation accounted for.
The correlation of +.419 indicated that 18 per cent of the
variability in yield was accounted for by its association with num-
ber of kernels per spikelet. Little importance can be attached to
this, however, since the correlation was not significant.
For prediction purposes, use may be made of the regression
equation given by
Y = y + b(x - x)
where Y = predicted yield.
y = observed mean yield.
x = number of kernels per spikelet.
Since y = 1619.3 -f- 63 = 25.703, x = 86.9 -*- 63 = 1.3794, and
b for variety regression was +11.109,
Y = 25.70 + 11.109(* - 1.3794).
Multiplying 1.3794 by 11.109 and adding 25.70 gives
Y = 10.38 + ll.WQx
From this equation, the predicted values of F (yield per acre)
can be calculated for varieties with different numbers of kernels
per spikelet. A few such are calculated for illustration.
CORRELATION AND REGRESSION
331
Average number
Yield per
acre, bu.
V&riety
spikelet (x)
Observed (y)
Predicted (7)
Marquis
1.47
24.6
26.7
Ceres
1.73
30.5
29.6
Hope
1.13
22.2
22.9
Ceres X Hope No. 1 ....
1.50
20.8
27.0
The observed and predicted yields are given merely as an
example of procedure. Unless correlation or regression is signifi-
cant and relatively high, it is apparent that prediction values will
not be very accurate.
MEANS AND DIFFERENCES OF CORRELATION COEFFICIENTS
Frequently it is desired to determine the significance of a
difference between two correlation coefficients. The method will
be illustrated with the use of the interannual correlation coeffi-
cients for loaf volume determined from grain of spring-wheat
varieties and strains grown in the regular rod-row nurseries in
four places in the state, as given by Ausemus et. al. (1938).
Since correlation coefficients cannot be averaged directly,
it is necessary to transform the coefficients to the statistic z
(Fisher 1938) and test the significance of the difference between
the z values by means of its error. The standard error of z is
The computations for testing the significance of the difference
between the interannual correlation coefficients for loaf volume
in 1929-1930, determined from 25 varieties, and in 1931-1932,
determined from 16 varieties, are carried through in Table 54.
TABLE 54. TEST OF SIGNIFICANCE OF A DIFFERENCE BETWEEN CORRELA-
TION COEFFICIENTS
Yea,rs correlated
Observed r
z
N - 3
Reciprocal
1929-1930
-f .43
.460
22
0455
1931-1932
4-. 15
.151
13
0769
Difference Sum .1224.
- .309
.360,
The observed values of r are first transformed to z with the
use of Appendix Table IV, The two correlation coefficients were
332
METHODS OF PLANT BREEDING
based on 25 and 16 pairs of observations each; so N 3 was 22
and 13, respectively. The sum of the reciprocals of JKT 3
is the variance of the difference between the values of z. The
square root of .1224 = .350 and is the standard error of the
difference. The difference was less than its standard error and,
therefore, it may be concluded that the two values of r were not
significantly different.
When several correlation coefficients for the same characters
are available, it frequently is desirable to determine the average
correlation. This may be done by transformations of r to z,
calculating the average value of z and then transforming the
average z back to r. With the use of data from the same study by
Ausemus et al., where the correlation coefficients of +.81, +.43,
and +.15 were based on 11, 25, and 16 determinations, respec-
tively, the calculations are carried through in Table 55.
TABLE 55. AVERAGING CORRELATION COEFFICIENTS
Years correlated
Observed r
z
,V - 3
(N - 3)z
1927-1928
4- 81
1 127
8
9 016
1929-1930 ...
4- 43
460
22
10 120
1931-1932
+ .15
.151
13
1 .963
4- 455
491
43
21.099
The values of r are first transformed to z, with the use of
Appendix Table IV. Each value of z is multiplied by N 3
and added to obtain 21.099. Dividing 21.099 by the sum of
N 3, or 43, gives .491 as the average value of z. This is then
transformed to r by means of Appendix Table IV to give an
average correlation coefficient of +.455. The standard error
of r = +.455 will be l/\/43 or .152. The accuracy of this
average correlation is equivalent to a single test involving
43 + 3 = 46 pairs of observations. This average correlation
is highly significant.
Before averaging correlation coefficients, it would be desirable
to test whether they are homogeneous, i.e., whether it can be
assumed that they could have arisen from a population with the
mean correlation coefficient through errors of random sampling.
The procedure in making such a test has been given by Rider
(1939). The computations are carried through in Table 56 7
with the use of data given in Table 55,
CORRELATION AND REGRESSION 333
TABLE 56. TEST FOB HOMOGENEITY OF COBBELATION COEFFICIENTS
Years correlated
r
z
N - 3
(N - 3)2
(N - 3)z 2
1927-1928
-f 81
1.127
8
9.016
10.161
1929-1930
+ .43
0.460
22
10.120
4.655
1931-1932 . . .
+ 15
151
13
1.963
0.296
Sum
43
21.099
15 112
Homogeneity of z can be tested by means of the % 2 test, where
X 2 = S(N - 3)z 2 -
for
- 15.112 -
S(N - 3) ~ "'"~ 43
k 1 = 2 degrees of freedom,
_ 4.759
where k = number of correlation coefficients.
In this problem, % 2 does not reach the 5 per cent point of 5.99
(Appendix Table III) for 2 degrees of freedom. It may be con-
cluded that these three correlation coefficients could have come
from equally correlated populations, the mean of which was
found previously to be r = +.455.
PARTIAL CORRELATION
An extension of the idea of correlation leads to its application to
groups of more than two variables. Partial- and multiple-
correlation coefficients then become of considerable interest.
Frequently two characters are related because of a third variable
that affects both. By means of partial correlation, the relation-
ship between two variables may be determined when the effect
of other variables is eliminated.
High-yielding varieties of grain of high quality are two of the
major objectives in crop improvement. Resistance to disease
is of major importance, also, if the disease affects the yield or
quality of the crop. In order to plan the breeding program, it
is necessary that the plant breeder have a knowledge of the
characters that are of greatest value under particular environ-
mental conditions and the interrelationships between them.
In determining the interrelationships between a number of
characters, the method of partial correlation is useful in deter-
mining the relationship between two characters independent of
the accompanying variation due to the other variables.
334
METHODS OF PLANT BREEDING
TABLE 57. MEAN YIELD, PLUMPNESS OF KERNEL, DATE HEADING, AND
CROWN-RUST PERCENTAGE IN ROD-ROW TRIALS WITH OATS
Variety or strain.
Nursery
stock
number
Yield
Plump-
ness
Date
heading
Crown
rust
Victory
514
36.5
3
7-11
14
Minota . ....
512
38
9
7-11
17
Miriota X White Russian
11-18-37
60.2
58
7-7
11
Black Mesdag
40.2
13
7-3
65
Double Cross ...
11-22-35
36.3
17
7-8
30
Double Cross
11-22-36
40.0
15
7-6
38
Double Cross .
11-22-37
51.8
43
7-6
25
Double Cross
11-22-38
57 3
28
7-7
10
Double Cross
11-22-39
40 6*
5
7-6
60
Double Cross . ....
11-22-40
49.0
12
7-5
60
Double Cross
11-22-41
43.8
7
7-5
57
Double Cross
11-22-42
39.4
7
7-5
60
Double Cross . .
11-22-43
48.5
13
7-4
40
Double Cross
11-22-44
40.7
2
7-6
50
Double Cross .
11-22-45
48.7
37
7-4
28
Double Cross .
11-22-46
51.0
28
7-5
20
Double Cross
11-22-47
40.8
5
7-5
40
Double Cross
11-22-48
38 5
7
7-7
33
Double Cross. ...
11-22-49
40.1
10
7-7
23
Double Cross
Double Cross
Double Cross
Double Cross
Double Cross
11-22-50
11-22-51
11-22-52
11-22-53
11-22-54
59.7
45.7
33.0
49 5
53.9
30
5
7
48
37
7-5
7-7
7-7
7-3
7-3
8
20
40
43
65
Double Cross
11-22-55
54 4*
50
7-3
63
Double Cross ....
Double Cross
Double Cross
11-22-56
11-22-57
11-22-58
37.2
40 5
48.8
32
25
32
7-4
7-5
7-3
50
38
60
Double Cross
11-22-59
47 6
15
7-4
53
Double Cross ....
11-22-60
51.1
23
7-4
50
Double Cross ... ...
11-22-61
53.4
23
7-4
53
Double Cross
Double Cross
11-22-62
11-22-63
55.9
54 . 9
52
55
7-4
7-4
27
18
Double Cross
11-22-64
46.2
15
7-3
47
Double Cross
11-22-65
49 3
10
7-3
30
Double Cross ...
11-22-66
46.4
35
7-3
33
Double Cross
11-22-67
54.4
23
7-4
35
Double Cross
11-22-68
52.1
23
7-5
20
Double Cross
11-22-69
70 5
67
7-3
15
Double Cross
11-22-70
72 9
67
7-3
25
Double Cross
11-22-71
21 2
7
7-7
40
Double Cross
11-22-72
24 6
7-10
30
11-22-73
53.2
57
7-3
37
Double Cross
11-22-74
50,9
17
7-5
37
Double Cross
11-22-75
61.7
30
7-5
15
11-22-76
53.4
12
7-7
12
Double Cross
11-22-77
43.1
22
7-4
25
Double Cross.
11-22-78
54.7
13
7-6
15
Double Cross. * . . . . ,
11-22-79
57.2
47
7-5
7
Double Cross
11-22-80
38 8
10
7-4
37
* Grown in two plots.
CORRELATION AND REGRESSION
335
An illustration of the methods of computation will be made
with data collected at University Farm, St. Paul, Minnesota,
from rod-row trials of oats, where three plots of each variety
or strain were grown and the average of the three replications
was used. Yield was expressed in bushels per acre, plumpness
of grain was a visual note taken as a percentage, and the amount
of crown rust was determined in percentage*
Plumpness of grain in small grains has been found to be
directly correlated with yield. Frequently yield is associated
with earliness. Both yield and plumpness of grain are influenced
to a considerable extent by rust. By means of partial correla-
tion, it was possible to determine the degree of association
between yield and plumpness when the effect of differences in
rust reaction was eliminated. Data to illustrate the computa-
tion of partial-correlation coefficients are given in Table 57.
For a more complete description of partial- and multiple-
correlation methods, the reader may be referred to Wallace and
Snedecor (1931).
In the methods to be given, the first step is the calculation of
the simple, or total, correlation coefficients. For convenience
of presentation, the following symbols will be used:
A = yield in bushel per acre.
B = plumpness of kernel.
C = date of heading.
D = percentage infection with crown rust.
The total correlation coefficients for all possible relationships
between these four variables are given in Table 58.
TABLE 58. TOTAL CORRELATION COEFFICIENTS FOR ALL INTERRELATION-
SHIPS BETWEEN YIELD, PLUMPNESS, DATE OF HEADING, AND PERCENT-
AGE OF CROWN RUST
A
B
C
B
4-. 7344*
C
-.4898*
-.4968*
D.
-.31951
- . 2320
.4012*
* Exceeds the 1 per cent level of significance,
t Exceeds the 5 per cent level of significance.
In this study, there were 50 pairs in the sample, and the degrees
of freedom for testing the significance of a total-correlation
coefficient would be N ~~ 2 or 48. If Appendix Table V is
336
METHODS OP PLANT BREEDING
referred to, it is noted that all correlation coefficients except
that between plumpness and crown-rust infection, TBD = .2320,
exceeded the 5 per cent point and that all but this coefficient and
r AD = .3195 exceeded the I per cent point.
A simple method of calculating partial-correlation coefficients
will be illustrated in detail. The partial-correlation coefficients
will be calculated from the standard parti al-regression coeffi-
cients by utilizing the fact that r 12.34 = \/fin-u X ^21-34, where
7*12. 34 means the correlation between variables 1 and 2 with 3 and 4
eliminated and /3i 2 . 34 and 21.34 are the standard regression coeffi-
cients. These regression coefficients are calculated by solving
sets of normal equations as illustrated in Table 59.
TABLE 59. SOLUTION OF NORMAL EQUATIONS TO OBTAIN STANDAKD
PARTIAL- REGRESSION ( ^EFFICIENTS
Line
D
C
B
A
Sum
Enter TDD roc, TDB, TDA D.
Change signs
1
2
1.0000
1 . 0000
- .4012
+ .4012
~ .2320
+ .2320
-.3195
+ .3195
+ .0473
3
1 . 0000
- .4968
. 4898
.3878
Multiply line 1 by 2.C
4
- .1610
.0931
. 1282
+ .0190
Add lines 3 and 4
Divide line 5 by 5.C, and change
signs
5
G
+ -8390
1 0000
- .5899
+ 7031
-.6180
+ 7366
- .3688
+ 4397 /
Enter TBS TBA. . . B,
7
1 0000
+ 7344
+ 1 0056
Multiply line 1 by 2.B . .
8
0538
0741
+ 0110
Multiply line 5 by 6.B
9
- .4148
.4345
. 2593
Add lines 7, 8, and 9 ...
10
+ .5314
+ . 2258
+ 7573
Divide line 10 by 10. B, and
11
- 1 . 0000
.4249
-1. 4251V
PAB.CD - +.4249
I
-h .4249
+ .4249
PAC.BD -.4379
II
- .4379
+ .2987
- . 7366
PAD.BC .3966
III
- .3966
- .1757
+ .0986
-.3195
NOTU: In the instructions 2. (7 represents +.4012 in line 2, column C, etc.
First the correlation coefficients are entered
added to obtain the sum. The correlation
obtain the sum for line 3, in the table, add the
coefficients in this line plus TCD. The sum for
of the total correlations in this line plus r B c and
directed in the instructions for each line, 1 to
The "sum" column serves as a check for all
in the table and
of TDD =1. To
three correlation
line 7 is the sum
TBD. Proceed as
11, in the table,
preceding work.
CORRELATION AND REGRESSION 337
The figures marked \/ must check, within rounding off of deci-
mals, the sum of the figures in that line to the left of the sum
column.
To calculate the partial-regression coefficients, bring down the
figures in column A, lines 11, G, and 2, in that order, and change
signs to form I, II, and III, column A. Write figure in LA one
column to the left. This is PAR- CD. Multiply I.fi by 6.fi and
2.Bj and write down products under II.fi and III.fi, respectively.
Thus, ( + .4249) X (+.7031) = +.2987 and (+.4249) X (+.2320)
= +.0986. Add II. A and II.fi to obtain ILC. This is the par-
tial-regression coefficient PAC.BD. Then multiply ILC by 2.C, or
(-.4379) X ( + .4012) = -.1757, and record as III.C. Add
III.A + III.fi + III.C to obtain III.D. This is the partial-
regression coefficient PAD.BC.
It is noted that in Table 59 the partial-regression coefficients
with A as a dependent variable were determined. The rule is
that the variable in the last column (if the column of sums is
omitted) is the first term of the regression coefficients; the second
term is that in the same vertical column. To obtain all possible
partial-regression coefficients, each variable must, in turn, be
placed in the last column and the set of normal equations solved
anew. To save time, it is best to set up the columns so that the
characters in the last two come in pairs, i.e., D, C, B, A ; D, C, A, fi
and A, B, D, C; A, B, C, D. By so doing, part of the computa-
tions from the first of a pair of characters can be copied off for
the second.
Through such calculations, &AB.OD = +.4249 and PBA.CD =
+ .5102. The partial-correlation coefficient will be
TAB. CD = VPAW.T/) X PBA.CD
= V^249"X^5l02 = +.4656
The significance of the partial-correlation coefficients may be
determined by reference to Appendix Table V for N 4 or
46 degrees of freedom. In general, the degrees of freedom are
N p 2, where N is the number of observations and p is the
number of variables eliminated. This amounts to the number of
observations minus the number of variables.
The more interesting partial-correlation coefficients are given
with the total-correlation coefficients for comparison.
338 METHODS OF PLANT BREEDING
TAB - +.7344* TAB-CD - +.4656*
TAC - -.4898* r^c.BD - -.4546*
TAD - -.3195t r^.^c - -.4600*
* Exceeds the 1 per cent point,
t Exceeds the 6 per cent point.
The partial correlation between yield and plumpness was highly
significant even after the effects of date heading and amount o
crown rust were eliminated. This strong relationship betweer
yield and plumpness of grain may be made use of during th<
segregating generations in a breeding program when yield tests
are not practical. Selection may be made for plumpness o
grain, if it is recognized that plumpness is strongly associatec
with yield. The association between yield and date heading was
of the same order of magnitude after plumpness and crown-rusl
differences were eliminated as when these were not. The partia
correlation between yield and percentage of crown rust was
highly significant when the effect of date heading and plumpness
of seed was eliminated. Apparently crown rust significantly
reduced yields apart from the influence of plumpness of grain
and date of heading.
MULTIPLE CORRELATION
The multiple-correlation coefficient measures the degree tc
which the dependent variable is influenced by a series of othei
factors studied. It may be calculated from the total-correlation
coefficients and the standard partial-regression coefficients. The
formula is
R 2 A-BCD = (TAB X $AB-CD) + (TAG X &AC.BD) + (TAD X &AD-BC)
By substituting the values of r and ft obtained in this problem
&A.BCD = (.7344 X .4249) + (-.4898 X -.4379)
+ (-.3195 X -.3966) = .6532
R = .8082
The significance of a multiple-correlation coefficient may be
tested by using Appendix Table V for N 4 = 46 degrees ol
freedom and entering the column for four variables. The
multiple-correlation of R = .8082 was highly significant. Squar-
ing this correlation coefficient gives 65 per cent as the percentage
of the variability in yield accounted for by its association with
plumpness of grain, date of heading, and percentage of crown rust,
CHAPTER XXII
MULTIPLE EXPERIMENTS, METHODS FOR TESTING A
LARGE NUMBER OF VARIETIES, AND THE ANALYSIS
OF DATA EXPRESSED AS PERCENTAGES
MULTIPLE EXPERIMENTS IN RANDOMIZED BLOCKS
The same 10 varieties of barley as those used in Chap. XX
to illustrate the computations for a randomized-block trial at
University Farm were tested also in five other stations in Minne-
sota, namely, Waseca, Morris, Crookston, Grand Rapids, and
Duluth. These tests were conducted in order to determine
varietal adaptation in different regions of the state. Since
studies of this nature will be of frequent occurrence in regional
trials, it will be of interest to illustrate the methods of analysis
that can be made.
In Table 60 is given the yields of 5 of the 10 varieties of barley
in each of 3 randomized blocks at each of 4 stations for each of
2 years. Since the computations are designed only to illustrate
the principles involved, the data are reduced from the more
extended test actually made. The analysis of the data follows
closely the method outlined by Immer, Hayes, and Powers (1934).
The degrees of freedom for an individual test at one location in
1 year would be keyed out as follows:
Variation Due to Degrees of Freedom
Blocks 2
Varieties 4
Error (blocks X varieties) 8
Total 14
For the complete analysis of all data combined, the degrees of
freedom could be keyed out as given in the summary on page 341 .
The key to the degrees of freedom on page 341 illustrates the
analogy between the individual tests and the complete analysis
for all data combined. It is to be noted that the degrees of
freedom for blocks within tests, varieties Within tests, and error
339
340
METHODS OF PLANT BREEDING
TABLE 60. YIELDS OF FIVE VARIETIES OP BARLEY, REPLICATED THREE
TIMES IN EACH OF FOUR LOCATIONS IN 1932 AND 1935
Variety
Block number
Block number
Sum
for
both
years
I
II
III
Sum
I
II
III
Sum
Manchuria . . .
Glabron ....
Velvet
University Farm, 1932
University Farm, 1935
236.5
248.4
251.4
304
239.9
19.7
28,6
20 3
27,9
22.3
31.4
38 3
27.5
40
30.8
29 6
43 5
32.6
46 1
31 1
80.7
110 4
80 4
114.0
84.2
45 5
47.5
54.2
62 2
47 4
50 3
41.1
52.3
53.1
57 8
60.0
49.4
64.5
74.7
50.5
155.8
138.0
171
190
155.7
Barbless
Peatland .
Sum
118.8
168
182.9
469.7
256.8
254 6
299 1
810 5
1280.2
Manchuria . . .
Glabron
Velvet
Waseca, 1932
Waseca, 1935
40.8
44 4
44.6
39 8
71 5
29.4
34.9
41.4
39 2
47 6
30 2
33.9
26 2
29 1
55 4
100.4
113.2
112.2
108 1
174.5
53 9
63.7
53.9
74.2
51.1
58 8
61.1
59 1
75 6
47 3
47 7
52.2
56.4
67.0
45
160.4
177
169.4
216.8
143 4
260.8
290.2
281.6
324 9
317.9
Barbless
Peatland
Sum
241.1
192 5
174 8
608.4
296.8
301 9
268.3
867
1475.4
Manchuria ....
Glabron
Crookston, 1932
Crookston, 1935
34.7
28.8
29 8
27.7
43
29.1
28.7
38 4
27 6
32 7
35.1
21.0
28.0
20.4
32.0
98.9
78.5
96 2
75.7
107.7
42.1
38.8
42.1
44.3
53.9
47.1
29.4
40.0
43 5
51 8
30 8
30.5
39.8
47 7
50 3
120.0
98 7
121.9
135.5
156
218.9
172.2
218.1
211.2
263.7
Velvet
Barbless
Peatland
Sum
164.0
156.5
136 5
457.0
221.2
211 8
199.1
632.1
1089.1
Manchuria ....
Glabron . .
Grand Rapids, 1932
Grand Rapids, 1935
20.2
13 2
24.5
19.0
27 6
30 2
20.5
41 6
18.4
30.0
16.0
9.6
30.6
24 6
22 7
66.4
43 3
96.7
62.0
80.3
26.6
21.4
20.7
20.7
32.6
26.5
18 7
26 8
23 6
40.0
32.7
24.1
30 4
30,9
34 2
85.8
64.2
77.9
75.2
106.8
152.2
107.5
174.6
137.2
187.1
Velvet
Barbless
Peatland
Sum
104.5
628.4
140.7
657.7
103.5
597.7
348.7
1883.8
122.0
896.8
135.6
903.9
152 3
918.8
409.9
2719.5
758.6
4603.3
Sum of 4
stations .
MULTIPLE EXPERIMENTS 341
are the product of the degrees of freedom in a single test multi-
plied by the number of tests.
Degrees of
Variation Due to Freedom
Between tests 7
Stations 3
Years 1
Stations X years 3
Between blocks within tests 16
Blocks 2
Blocks X stations 6
Blocks X years 2
Blocks X stations X years 6
Between varieties within tests 32
Varieties 4
Varieties X stations 12
Varieties X years 4
Varieties X stations X years 12
Error within tests 64
Blocks X varieties 8
Blocks X varieties X stations 24
Blocks X varieties X years 8
Blocks X varieties X stations X years 24
Total Tl9
The degrees of freedom for the main effects, such as stations,
years, blocks, and varieties, are 1 less than the number of sta-
tions, years, blocks, or varieties, respectively. The degrees of
freedom for the interactions are the product of the degrees of
freedom for the main effects involved.
Once the degrees of freedom are keyed out, the calculation of
the sums of squares must be made in accordance with this plan.
Before proceeding with the complete analysis, it will be well to
test the errors of the eight separate tests for homogeneity in
order to .determine whether they may legitimately be combined
into a single analysis with a single error. In Table 61 are given
the sums of squares calculated separately for each of the eight
tests.
The % 2 (Chi-square) distribution can be used as an approximate
test of the homogeneity of several estimates of variance. The
method proposed by Bartlett (1937) will be used to determine
whether the variances calculated for the separate tests can be
considered homogeneous, i.e., whether they can be considered
random sampling deviates from the mean of these variances.
342
METHODS OF PLANT BREEDING
The formula for x 2 will be x 2 = -^ [n log e s 2 S(n r log c s*)} for
o
fc 1 degrees of freedom, where k is the number of variances
being compared, n r is the degrees of freedom of each variance,
n is the total degrees of freedom for the separate variances and
TABLE 61. SUMS OP SQUARES CALCULATED FOR THE SEPARATE TESTS
Station and year of test
Sums of squares for
Total
Blocks
Varieties
Error
University Farm, 1932
University Farm, 1935...
Waseca 1932
867 30
1031 71
1907 14
1203.56
487 87
807 84
905 56
509 91
450 10
251 62
471.40
131 15
80 83
49 21
179.69
92 13
375 . 61
506 36
1196 33
993.84
252 62
595.82
536.17
336 46
41 59
273 73
239 41
78.57
154 42
162.81
189.70
81.32
Waseca, 1935
Crookstoii, 1932
Crookston 1935
Grand Rapids, 1932
Grand Rapids, 1935
Sura
7720 89
1706 13
4793 21
1221.55
TABLE 62. CALCULATION OF % 2 TEST FOR HOMOGENEITY OF VARIANCES
Degrees
Station and year
of test
of free-
dom in
Lrror
variance
of each
log, si
n r sl
n r log, s 2 r
each
test, s 2
test, n r
University Farm, 1932
8
5.20
1 6487
University Farm, 1935
8
34 22
3 5328
Waseca, 1932
8
29 93
3 3989
Waseca, 1935
8
9.82
2 2844
Crookston, 1932
8
19.30
2 9601
Crookston, 1935
8
20.35
3.0131
Grand Rapids, 1932. .
8
23.71
3 1659
Grand Rapids, 1935. .
8
10 16
2 3184
Sum
64
152.69
22 . 3223
1221.52
178 5784
S(n r ), s? refers to the individual variances, s 2 the pooled variance
calculated from S(n r s%)/n, and C is a correction term denned by
C= 1 +
3(*
=D \ a Q-J -
MULTIPLE EXPERIMENTS 343
The estimated variances will be found by dividing the error
sums of squares in Table 61 by 8 degrees of freedom. Table 62
may be formed to illustrate the computations.
Since the degrees of freedom are the same for each test, the
sum of the products S(n r s*)j and S(n r log e s 2 -) ma 7 be calculated
from the totals of columns 2, 3, and 4 in Table 62; otherwise each
value of n r sl and n r log c s? would need to be calculated and the
column added to obtain the sum. In this problem S(n r $l) =
1221.52 is found by multiplying 152.69 by 8 and S(n r log e s?) =
178.5784 is obtained by multiplying 22.3223 by 8. Since
n = 64
s2 = 122L52
t>4
n log, s 2 = 64 X 2.94891 = 188.7302
c = i + Hi { (M + H + M + H + H + y s + y s + M)
- 1^4! = 1-0469
. 188.7302 - 178.5784
X = - - = 9.70, tor? degrees ot ireedom.
-L .
Referring to the table of % 2 (Appendix Table III) for 7 degrees of
freedom, we find that x 2 " 9.80 when P = .20. We may con-
clude, therefore, that deviations between variances as great as
these observed would occur more than 20 times in 100 through
errors of random sampling. These 8 error variances may, there-
fore, be considered homogeneous, and it will be legitimate to
replace the 8 separate variances by their mean variance (19.09)
in the analysis of variance of all data.
The total yield of the 120 plots, as given in Table 60, was
_ noo , ^ ... , ,_ [S(x)] 2 (4603.3) 2
4603.3 bu. The correction term S(x)x = AT - = ~ =
.A/ 1^0
176,586.42. To obtain the total sum of squares, the squares of
the individual plots are added to give S(x) 2 = 200,879.35, and
the correction term S(x)x is subtracted to give 24,292.93.
Several other tables need to be set up by adding the appropriate
yields in Table 60. In Table 63 are given the total yields for
each variety at each station by adding the yields for both years.
The figures in Table 63 are taken directly from the right-hand
margin of Table 60. They are assembled here for convenience,
with the appropriate variety and station totals,
344
METHODS OF PLANT BREEDING
In Table 64 are given the data for comparisons of varieties in
different years, obtained by adding the yields at the four stations.
TABLE 63. TOTAL YIELDS GROUPED FOR VARIETIES AND STATIONS
Variety
Station.
Sum
University
Farm
Waseca
Crookston
Grand
Rapids
Manchuria, . . .
236.5
248.4
251.4
304.0
239 9
260.8
290.2
281 6
324.9
317 9
218 9
177 2
218.1
211.2
263 7
152 2
107.5
174.6
137.2
187 1
868.4
823.3
925.7
977 3
1008 6
Glabroii
Velvet
Barbless . .
Peatland
Sum
1280 2
1475 4
1089 1
758 6
4603 3
TABLE 64. TOTAL YIELDS GROUPED FOR VARIETIES AND YEARS
Variety
Year
Sum
1932
1935
Manchuria
346 4
345 4
385 5
359 8
446 7
522.0
477.9
540.2
617.5
561.9
868.4
823 3
925.7
977 3
1008 6
Glabron . ...
Velvet
Barbless . . .
Peatland
Sum ... . ...
1883 8
2719.5
4603 3
In Table 65 are assembled the data- for comparisons of blocks
and stations, obtained by adding the block totals for the 2 years
of each station.
TABLE 65. TOTAL YIELDS OF BLOCKS AND STATIONS
Block
Station
Sum
University
Farm
Waseca
Crookston
Grand
Rapids
I
375.6
422.6
482.0
537.9
494.4
443.1
385.2
368 3
335.6
226 5
276.3
255.8
1525.2
1561.6
1516.5
II
Ill
Sum
1280.2
1475 4
1089 1
758.6
4603.3
MULTIPLE EXPERIMENTS
345
In Table 66 are the totals for comparison of blocks and years.
This table is assembled from the totals at the bottom of Table 60.
TABLE 66. TOTAL YIELDS OF BLOCKS AND YEARS
Tllrknlr
Yc
jar
Qnrr
1932
1935
I
628 4
896 8
1525 2
II
657.7
903.9
1561 6
Ill
597 7
918 8
1516.5
Sum
1883.8
2719.5
4603 3
One other table is necessary, that of stations and years. This
is given as Table 67. The figures for this comparison are assem-
bled here for convenience, also, but could have been found
directly in Table 60.
TABLE 67. TOTAL YIELDS OF STATIONS AND YEARS
Year
Station
Sum
University
Farm
Waseca
Crooks ton
Grand
Rapids
1932
469.7
810.5
608.4
867.0
457.0
632.1
348.7
409 9
1883.8
2719.5
1935 ....
Sum
1280.2
1475.4
1089.1
758.6
4603 3
The calculation of the sums of squares for the complete analysis
can be performed with the least difficulty and confusion if the
steps are carried through in a routine manner. Many of the
calculations are given in Table 68. The remainder follow easily
and logically.
In Table 68, x is used to designate the individual plots and
x 8 , x v , x vj and Xb, the total yields for each station, year, variety,
and block, respectively. The symbols x V8 , x vy , etc., refer to the
totals for each variety at each station, each variety each year,
etc., as found within Tables 63 to 67. The symbol x vsy refers to
the total yield of each variety at each station for a single year.
Column 2 of Table 68 gives the number of figures squared in
calculating column 1, and column 3 gives the number of plots
346
METHODS OF PLANT BREEDING
in each figure squared. Column 4 is necessary to reduce the
sums of squares to a single-plot basis. Column 6 gives the sums
of squares, and column 7 the degrees of freedom.
TABLE 68. CALCULATION OF SUMS OF SQUARES
Num-
Variate
Total of
squares
Num-
ber of
figures
squar-
ber of
plots
in each
figure
Correc-
tion
term
S(x)x
Sum of
squares
De-
grees
of free-
dom
ed
squar-
ed
(4 =
(6 -
(1)
(2)
(3)
1 -5- 3)
(5)
4-5)
(7)
8(x*)
200,879.35
120
1
200,879.35
176,586.42
24,242.93
119
S(xl)
5,577,329.97
4
30
185,911.00
176,586.42
9,324.58
3
S(xl)
10,944,382.69
2
60
182,406.38
176,586.42
5,819.96
1
8(xl)
4,261,251.19
5
24
177,552.13
176,586.42
965.71
4
S(x%)
7,064,601.85
3
40
176,615.05
176,586.42
28.63
2
S(xl a )
1,129,020.73
20
6
188,170.12
176,586.42
11,583.70
19
S(xl v )
2,206,627.61
10
12
183,885.63
176,586.42
7,299.21
9
S(xl s )
1,871,824.37
12
10
187,182.44
176,586.42
10,596.02
11
S(xl y )
3,650,180.03
6
20
182,509.00
176,586.42
5,922.58
5
S(x' 2 8y )
2,897,377.01
8
15
193,158.47
176,586.42
16,572.05
7
S(xl av )
974,322.93
24
5
194,864.59
176,586.42
18,278.17
23
S(xl 8V )
593,855.03
40
3
197,951.68
176,586.42
21,365.26
39
The sums of squares for the first-order interactions are obtained
by subtracting from the sum of squares for the two variables in
Table 68 the sums of squares for the two main effects. For
example, the sum of squares for the interaction of varieties X sta-
tions will be given by the sum of squares for x\ 9 in Table 68
minus the sum of squares for the main effects of varieties and
stations x\ and x%. Numerically, this will be:
Sum of Squares
11,583.70
- 965.71
- 9,324.58
Degrees of Freedom
19
4 (varieties)
3 (stations)
1,293 41 12 (varieties X stations)
The second-order interaction of varieties X stations X years,
for example, is obtained by subtracting from the sum of squares
opposite S(xl 8y ) in Table 68 the sums of squares for varieties,
stations, years, and all possible first-order interactions. Thus:
MULTIPLE EXPERIMENTS
347
Sum of Squares
21,365.26
- 965.71
- 9,324.58
- 5,819.96
- 1,293.41
- 513 54
- 3,427.51
2,020.55
Degrees of Freedom
39
4 (varieties)
3 (stations)
1 (years)
12 (varieties X stations)
4 (varieties X years)
3 (stations X years)
12 (varieties X stations X years)
The complete analysis of variance is now carried through in
Table 69, the error sum of squares being obtained as a remainder.
TABLE 69. COMPLETE ANALYSIS OF VARIANCE
Variation due to
Degrees
of freedom
Sum of
squares
Mean
square
s
F
Stations
3
1
3
2
6
2
6
4
12
4
12
64
9,324 58
5,819.96
1,427 51
28.63
1,242 81
73 99
360.69
965 71
1,293 41
513 54
2,020 55
1,221 55
3108 19
5819 96
475.84
14 32
207 14
37.00
60 12
241 43
107.78
128 39
168 38
19.09
4.37
162.82*
304.87*
24.93*
10 85*
1 94
3.15*
12.65*
5.65*
6.73*
8.82*
Years
Stations X years
Blocks
Blocks X stations ... .....
Blocks X years
Blocks X stations X years ....
Varieties
Varieties X stations
Varieties X years
Varieties X stations X years.
Error
Total
119
24,292.93
* Exceeds the 1 per cent level of significance when compared with error mean square.
The structure of the complete analysis of variance becomes
clear when it is compared with the separate analyses of variance
of the single tests. The sum of squares for error in Table 69 is
the same as for the sum of all tests calculated separately in
Table 61. The error mean square in the complete analysis is,
then, the average of the eight individual error mean squares
calculated separately in Table 62. Thus, 152.69 -5- 8 = 19.09,
the mean square for error given in Table 62. Adding the sum
of squares for varieties, varieties X stations, varieties X years,
and varieties X stations X years in Table 69 gives 4793.21.
This agrees with 4793.21 obtained as the sum of the sums of
squares for varieties within tests given in Table 61. A similar
comparison holds for blocks.
348
METHODS OF PLANT BREEDING
The manner in which the data can be interpreted will now be
illustrated. From Table 69, it is seen that the mean square for
varieties, varieties X stations, varieties X years, and varie-
ties X stations X years, compared with error mean square,
exceeded the 1 per cent point. It is plain, therefore, that there
were significant differences in average yielding ability and that
some varieties reacted in a differential manner at some stations
and in some years.
A summary of the mean yields of the five varieties for 1932
and 1935 at each of the four stations is given in Table 70. The
varieties are listed in the order of average yield at all stations.
TABLE 70. MEAN YIELD OF FIVE VARIETIES OF BAKLEY FOB 1932 AND 1935
AT EACH OF FOUR STATIONS AND THEIR AVERAGE YIELD AT ALL
STATIONS
Variety
Station
Average
University
Farm
Waseea
Crookston
Grand
Rapids
Peatland
40.0
50.7
41.9
39.4
41.4
53.0
54.2
46.9
43.5
48.4
44.0
35.2
36.4
36.5
29.5
31.2
22.9
29.1
25.4
17.9
42.0
40.7
38.6
36.2
34.3
Barbless
Velvet
Manchuria
Glabron
The standard error of a single plot (Table 69) was 4.37 bu.
Since 24 plots were involved in the variety averages for all sta-
tions and both years, the standard error of the mean of 24 plots
would be 4.37/\/24 = 0.89 bu., and the standard error of a differ-
ence between two variety means would be 0.89 \/2 = 1.26 bu.
A formula frequently used to obtain the standard error of an
average of several tests is 1/JV" \/sl 4~ s| + &c + * " " > where N
is the number of tests and sj, s|, etc., are the variances for error
of the separate tests. If the variances of the mean of three plots
were calculated for each of the eight tests by dividing each of the
error variances, given as si in Table 62, by 3, the standard error
of the average of all tests would be
i.20 + 34.22 +
10.16
= 0.89 bu., the same as calcu-*
lated in the preceding paragraph.
MULTIPLE EXPERIMENTS 349
With 64 degrees of freedom for error, t 2 at the 5 per cent
level of significance. If twice the standard error of the difference
between two means as a minimum level of significance is accepted,
it may be said that differences in excess of 2 X 1.26 = 2.52 bu.
would be judged as probably significant. On this basis, Velvet,
Manchuria, and Glabron would be significantly lower in yield
than Peatland. Manchuria and Glabron would be significantly
lower in yield than Barbless. Manchuria and Glabron were
lowest in yield, and the difference between them was not
significant.
The mean square for varieties X stations was significantly
greater than the mean square for error. It is apparent that some
varieties reacted in a differential manner at certain stations, A
first-order interaction involves the difference between two differ-
ences. The mean yield of Barbless at University Farm, for
an average of both years, exceeded that of Peatland by 50.7
40.0 = 10.7 bu. per acre. At Grand Rapids, however, the differ-
ence between these two varieties was 31.2 22.9 = 8.3 bu. in
favor of Peatland. The question arises whether these two differ-
ences are significantly different. This difference between two
differences will be (50.7 - 40.0) - (22.9 - 31.2) = 19.0 bu.
Since six plots were involved in each mean being compared, the
standard error of the cross difference will be ^ \/2 \/2 = 3.57
V6
bu. Twice this is 7.14 bu., and any "cross difference " exceeding
this value is*expected to occur less than once in 20 trials by
random sampling alone. It is clear, therefore, that the yields of
the varieties Barbless and Peatland were differential at Uni-
versity Farm and Grand Rapids. Other differential responses
can be found also in Table 70. Extensive testing of these and
other standard varieties of barley in Minnesota for a long period
of years has shown Peatland to be better adapted at Grand
Rapids than at any other experiment station in the state. From
data such as these, carried out at six stations in the state for a
minimum period of 3 years, general recommendations regarding
varieties are made to the farmers. In many instances, significant
interactions of varieties and stations are obtained, and certain
varieties are recommended only for certain- regions of the state.
Significant interactions of some varieties in the 2 years could
be found by applicatioB of the general procedure outli&ed above.
350 METHODS OF PLANT BREEDING
Interactions of varieties X years are of less interest to the plant
breeder than interactions of varieties X stations.
Although the second-order interaction of varieties X stations X
years was significant also, this is of minor interest to the plant
breeder. A significant second-order interaction means that cer-
tain differential responses of two varieties at each of two stations
was not the same in each of 2 years.
For a complete understanding of an analysis of variance, of
which that given in Table 69 is an example, one further com-
parison can be set up. Letting V, S 7 and Y represent variances
due to varieties, stations, and years, respectively, and V X S,
V X Y, and V X S X Y the interaction variances, we may deter-
mine whether variance due to
V>VXS>VXSXY
yerror
V>VXY>VXSXY 7
by means of the F test. The symbol > means "greater than."
If variance due to varieties significantly exceeds the interaction
of varieties X stations, we have evidence that varietal perform-
ance generally was consistent enough to demonstrate that some
varieties were the best in all stations, as an average of the years
in which tests were made. If the variety variance significantly
exceeds that of varieties X years, we may conclude that as an
average of all stations some varieties were consistently better in
yield in all years.
Further, if the interaction of varieties X stations significantly
exceeds varieties X stations X years, it is plain that the differ-
ential responses of the varieties at the separate stations were
sufficiently similar in the different years to warrant the conclusion
that these differential responses may be permanent features of
these localities.
Unless the variance for varieties significantly exceeds that of
varieties X stations, no general recommendation of a variety for
the entire state can be made. Extensive tests in the region in
which the varieties may be grown provide the only sound basis
for recommendation over wide areas.
COMPARISON OF VARIETIES IN DIFFERENT EXPERIMENTS,
WHERE THE SAME CHECK VARIETIES ARE GROWN
Frequently it may be desirable to compare the yields of varie-
ties grown in different experiments, If the samQ check varieties
SIMPLE LATTICE EXPERIMENTS 351
have been included in the different experiments, comparisons of
the new varieties may be made by comparing them through the
checks. Thus, if A and B are the yields of two varieties in
different experiments and the same check (cK) has been included
in each test, the relative difference in yield between A and B will
be given by (A chi) (B c/i 2 ), where chi and c/i 2 are the
yields of the checks in the experiments involving A and J9,
respectively.
The standard error of the foregoing difference will be
\/2s\ + 2sf, where s\ and s\ are the variance of the mean for the
two tests.
If more than one check variety has been included in each test,
comparisons may be made of (A - ch\) (B ch%), where chi
and c/?2 are the means of the several checks. The standard error
of this difference will be
where sf and sf = variance of the mean for a single variety in
experiments 1 and 2.
N = number of checks used in each experiment.
SIMPLE LATTICE EXPERIMENTS
When the number of varieties to be tested is small, the random-
ized block or Latin-square designs provide an efficient method for
testing the significance of varietal differences. As the number
of varieties becomes large, the randornized-block design with all
varieties in the same block becomes less efficient because of
increasing soil heterogeneity within blocks. The Latin-square
design for large numbers of varieties cannot be used because of
the prohibitive number of replications required.
Yates (1936) suggested a modification of the complete block
design whereby the number of varieties in a block was less than
the total number to be tested. The error could then be calcu-
lated from the variation within the small, or incomplete, blocks
and would be lower, usually, than the error calculated from
randomized complete blocks. The methods of analysis appropri-
ate for such designs have been given by Yates (1936) and Goulden
(1937, 1939). In these analyses some information about varietal
difference was lost. As a result, these incomplete block designs
352
METHODS OF PLANT BREEDING
could be less efficient than ordinary randomized complete blocks
if the soil were relatively homogeneous. Recently Yates (1939)
and Cox, Eckhardt, and Cochran (1940) have shown how all the
information regarding differences between varieties in different
incomplete blocks could be recovered. As a consequence, these
designs can never be appreciably less efficient than ordinary
randomized blocks containing the total number of varieties and
will be considerably more accurate if there is a reduction in the
error through the use of the small, or incomplete, blocks.
TABLE 71. RANDOM; ARRANGEMENT OF VARIETIES IN LATTICE EXPERIMENT
Replication 1 (Group X) Replication 2 (Group Y)
Block Block
(1)
10
7
6
8
9
(2)
14
13
11
15
12
(3)
2
4
5
3
1
(4)
25
24
23
21
22
(5)
18
16
17
20
19
(6)
15
5
10
20
25
(7)
16
6
21
11
1
(8)
2
17
7
22
12
(9)
23
3
13
18
8
(10)
24
4
14
19
9
Replication 3 (Group X)
Block
Replication 4 (Group Y)
Block
(11)
8
9
10
6
7
(12)
23
21
24
25
22
(13)
12
14
11
13
15
(14)
16
19
20
17
18
(15)
3
4
5
1
2
(16)
13
8
3
23
18
(17)
2
22
12
17
7
(18)
19
14
9
24
4
(19)
21
11
16
1
6
(20)
10
15
20
5
25
The design and computation of the data for a lattice 1 experi-
ment will be illustrated using uniformity trial data given by
Wiebe (1935). It was assumed that 25 varieties grown in three-
row plots and the central row harvested were to be tested, with
four replications. The method of computation will follow
closely that given by Cox, Eckhardt, and Cochran (1940).
In the lattice experiments described here, the number of
varieties is a perfect square. The number of varieties, v = Jfc 2 ,
are tested in incomplete blocks of k varieties each. The varieties
1 Certain of the lattice designs have been referred to as pseudofactorial
arrangements in two equal groups of sets, as two-dimensional pseudofactorial
arrangements with two equal groups of sets, and as two-dimensional quasi-
factorial desigp in randomized blocks in two equal groups of sets.
SIMPLE LATTICE EXPERIMENTS
353
may be identified by numbers arranged in a square, as follows,
with the use of k z = 25 varieties :
12345
6 7 8 9 10
11 12 13 14 15
16 17 18 19 20
21 22 23 24 25
TABLE 72. YIELDS OF VARIETIES IN GRAMS PER ROD Row
Replication 1 (Group X)
Block Block totals
1
2
3
4
5
(3)
635
525
555
650
635
3,000
6
7
8
9
10
(1)
495
730
810
775
710
3,520
11
12
13
14
15
(2)
630
600
645
635
645
3,155
16
17
18
19
20
(5)
735
690
840
855
805
3,925
21
22
23
24
25
(4)
620
795
590
660
615
3,280
16,880
Block
Replication 2 (Group Y)
Block totals
1
6
11
16
21
(7)
530
490
595
495
540
2,650
2
7
12
17
22
(8)
610
660
620
695
570
3,155
3
8
13
18
23
(9)
705
850
675
685
640
3,555
4
9
14
19
24
(10)
840
905
785
860
875
4,265
5
10
15
20
25
(6)
670
455
655
665
615
3,060
16,685
354
METHODS OF PLANT BREEDING
TABLE 72. YIELDS OF VARIETIES IN GRAMS PER HOD Row. (Continued)
Replication 3 (Group X)
Block Block totals
1
2
3
4
5
(15)
635
700
640
640
645
3,260
6
7
8
9
10
(ID
570
545
675
580
470
2,840
11
12
13
14
15
(13)
550
515
450
550
495
2,560
16
17
18
19
20
(14)
505
620
700
570
575
2,970
21
22
23
24
25
(12)
445
455
445
465
515
2,325
13,955
Block
Replication 4 (Group F)
Block totals
1
6
11
16
21
(19)
550
655
545
515
550
2,815
2
7
12
17
22
(17)
455
425
460
470
440
2,250
3
8
13
18
23
(16)
445
545
700
530
525
2,745
4
9
14
19
24
(18)
510
525
515
395
425
2,370
5
10
15
20
25
(20)
540
575
610
510
615
2,850
13,030
These 25 varieties are arranged in incomplete blocks of k = 5
varieties each. In one group, designated as J, the 5 varieties
for each block are taken from a row of the foregoing square. In a
second group, designated F, the varieties are taken from a column
of the foregoing square. The order of the 5 rows, or columns, is
randomized, and the 5 varieties within each row or column are
SIMPLE LATTICE EXPERIMENTS
355
TABLE 73. COMBINATION OF REPLICATIONS
Group X (Replication 1 + Replication 3)
Row totals
1
2
3
4
5
1270
1225
1195
1290
1280
6,260
6
7
8
9
10
1065
1275
1485
1355
1180
6,360
11
12
13
14
15
1180
1115
1095
1185
1140
5,715
16
17
18
19
20
1240
1310
1540
1425
1380
6,895
21
22
23
24
25
1065
1250
1035
1125
1130
5,605
Column totals .
5820
6175
6350
6380
6110
30,835
Group Y (Replication 2 -f Replication 4)
Row totals
1
1080
6
1145
11
1140
16
1010
21
1090
5,465
2
1065
7
1085
12
1080
17
1165
22
1010
5,405
3
1150
8
1395
13
1375
18
1215
23
1165
6,300
4
1350
9
1430
14
1300
19
1255
24
1300
6,635
5
1210
10
1030
15
1265
20
1175
25
1230
5,910
Column totals
5855
6085
6160
5820
5795
29,715
placed in random order. In Table 71 on page 352 is given the
random arrangement of the 25 varieties in each of two replica-
tions of the X and Y groups.
Varieties 1, 2, 3, 4, and 5 fell in block 3 and again in block 15,
both in group X. Varieties 1, 6, 11, 16, and 21 fell in blocks 7
and 19 in group F. All 25 varieties are contained in each com-
plete replication.
356
METHODS OF PLANT BREEDING
The foregoing random arrangement of varieties was super-
imposed on the plot yields, in grams per rod row, of Wiebe's
(1935) data. The plot yields of each variety are given in
Table 72, assembled according to rows and columns of the original
square. The variety number is given above each plot yield.
The data from both replications of groups X and F, for each
variety, are added next and are given in Table 73.
The sums of the yields of the two plots each for the X and Y
groups are combined next to give the total yields of the four plots
of each variety. These total yields of the varieties are given in
Table 74, with appropriate row and column totals.
TABLE 74. TOTAL YIELDS OF VARIETIES
Row totals
1
2
3
4
5
2,350
2,290
2,345
2,640
2,490
12,115
6
7
8
9
10
2,210
2,360
2,880
2,785
2,210
12,445
11
12
13
14
15
2,320
2,195
2,470
2,485
2,405
11,875
16
17
18
19
20
2,250
2,475
2,755
2,680
2,555
12,715
21
22
23
24
25
2,155
2,260
2,200
2,425
2,360
11,400
Column totals
11,285
11,580
12,650
13,015
12,020
60,550
The total sum of squares is found by adding the squares of
the 100 plot yields in Table 72 to give 38,012,350 and subtracting
the correction term (60,550) 2 /100 = 36,663,025 to give 1,349,325
as the total sum of squares.
The sum of squares for replications is calculated from the totals
of the four complete replications, i.e.,
(16,880)' + (16,685) 2 + (13,955)' + (13,030)" _ 36)663)025
2t&
= 450,837
The sum of squares for varieties (ignoring blocks) is calculated
the variety totals in Table 74. Thus
(2350)* + (2290)' + + (2360)
SIMPLE LATTICE EXPERIMENTS
357
The sum of squares for blocks (eliminating varieties) is made
up of two components as follows:
Component a is calculated from the sum of squares of differ-
ences between blocks containing identical varieties. These differ-
ences are found by subtracting the block totals for the same
group of varieties in replications 1 and 3 and in 2 and 4, taken
from Table 72. These differences are given below.
Set X
Set F
Replica-
Replica-
Differ-
Replica-
Replica-
Differ-
tion 1
tion 3
ence
tion 2
tion 4
ence
3,000
3,260
- 260
2,650
2,815
- 165
3,520
2,840
680
3,155
2,250
905
3,155
2,560
595
3,555
2,745
810
3,925
2,970
955
4,265
2,370
1895
3,280
2,325
955
3,060
2,850
210
16,880
13,955
2925
16,685
13,030
3655
The sum of squares of the deviations within these two sets of
differences will give the sum of squares between paired blocks
within sets. Thus
(-260) 2 +
(955) 2 + (-165) 2
(210) 2
10
_ (2925)* + (3655)' =
The divisors are 2k = 10 and 2k 2 = 50. 1
Component b is obtained from two sets of differences giving
estimates of block yields freed of varietal effects. In Table 73,
1 For six replications, there would be three columns of block totals for
each set. Component a would then be calculated from an analysis of
variance for each set, the degrees of freedom for set X being as follows:
Degrees
of
freedom
Replications 2
Set X totals k - 1
Interaction, or component (a) 2(k 1)
The same would be done for set F, and the degrees of freedom and sums of
squares of both added to give the complete component a.
358 METHODS OF PLANT BREEDING
the row totals are the sums of two blocks of the same group of
varieties. These row totals cannot be used to calculate block
sum of squares, since they are confounded with (contain) varietal
effects as well. The first row total in group X (Table 73), made
up of the sums of the two blocks containing varieties 1, 2, 3, 4,
and 5, comes to 6260. The first column total in group Y is 5855
and is clearly an estimate of the varietal effects alone of the
same five varieties, since each block in group F is equally repre-
sented. The difference between these totals, 6260 5855 = 405,
is an estimate of block effect freed of varietal differences. In a
similar manner, the first row total in group F (Table 73) minus
the first column total in group X, 5465 5820 = -355, is an
estimate of block effect for blocks containing varieties 1, 6, 11,
16, and 21.
Since, however, it is easier to add than subtract in making
adjustments to the average yields of the varieties, it is preferable
to subtract the un confounded from the confounded totals and
work with the negative values. Thus, 5855 6260 = 405
and 5820 5465 = 355. These differences are designated rkc x
and rkCy, where r is the number of replications and k the number
of plots per block and c x and c y the mean corrections from the
X and F groups, respectively. The rkc x values may be deter-
mined also by subtracting from the row totals of Table 74 twice
the row totals of group X in Table 73. Thus, 12, 1 15 - 2(6260) =
405. The rkc y values are obtained by subtracting from the
column totals in Table 74 twice the row totals for group F in
Table 73. These values are given below:
rkc x rkcy
12,115 - 2(6260) - 405 11,285 - 2(5465) = 355
12,445 - 2(6360) = - 275 11,580 - 2(5405) - 770
11,875 - 2(5715) = 445 12,650 - 2(6300) = - 50
12,715 - 2(6895) - -1075 13,015 - 2(6635) = -255
11,400 - 2(5605) = 190 12,020 - 2(5910) - 200
-1120 1120
The sum of the rkc x and rkc y values will be 0.
The sum of squares of deviations within sets of rkc x and rkc y
will be an estimate of variance between blocks (eliminating
varieties). Thus
SIMPLE LATTICE EXPERIMENTS
(-405)'+ + (190) 2 + (355) 2 + + (200) 2
359
20
QQ
(1120)' _
- 97,705
The divisors will be rk = 20 and rk 2 = 100.
The analysis of variance table may now be set up. This is
given in Table 75.
TABLE 75. ANALYSIS OF VARIANCE OF LATTICE EXPERIMENT
Variation due to
Degrees of
freedom
Sum of squares
Mean square
Replications . .
3
16
24
56
450,837
150,279 00
43,282.75
12,213.12
27,747.94
10,027.62
3,818.89
Component a
8
8
346,262
97,705
Component b .
Blocks (eliminating
varieties)
443,967
240,663
213,858
Varieties (ignoring
blocks)
Error (intrablock) . .
Total
99
1,349,325
A test of significance of variety mean square in the form of
an F test cannot be made from the mean squares for varieties
and error in Table 75, since variety mean square is partially
confounded, i.e.j it contains some of the differences between
blocks. Usually it will be sufficient to apply the test of signifi-
cance appropriate for an ordinary randomized-complete-block
test. Although less precise than the exact test, this usually will
be adequate. A lattice experiment can always be analyzed as
an ordinary randomized-complete-block test. Such a test is
given in Table 76.
TABLE 76. ANALYSIS OF VARIANCE AS RANDOMIZED COMPLETE BLOCKS
Variation due to
Degrees of
freedom
Sum of
squares
Mean
square
F
Replications
3
450,837
150,279.00
Varieties
24
240,663
10,027.62
1.10
Error
72
657,825
9,136.46
Total
99
1,349,325
360 METHODS OF PLANT BREEDING
In the foregoing table, the degrees of freedom and sums of
squares for replications, varieties, and total are taken directly
from Table 75 and the error degrees of freedom and sum of
squares obtained by subtraction. In this case, F = 10,027.62 -f-
9,136.46 = 1.10, a nonsignificant value for n\ = 24and^ 2 = 72
degrees of freedom, since uniformity trial data were used. If
significance of differences between variety means is indicated by
the test of significance applied to the randomized-complete-block
analysis, no further test of significance is necessary. Usually,
when large numbers of varieties are used, it may be expected
that significant differences between varieties will be found if the
conditions of the test have been satisfactory.
The average yield of the varieties, calculated from Table 74, is
affected by differences in productivity of the blocks. The neces-
sary corrections are obtained by weighting c x and c v to give c x >
and Cy'. These corrections are then added to the arithmetic
averages to give the adjusted mean yields.
The weighting factor is (w w')/(w + w f ), where w = \/E
and w' = 3/(4B E), E and B being, respectively, the error
and block mean squares in Table 75. If B is less than or equal
to E, no adjustments for blocks are necessary, and the averages
in Table 77 are the correct variety means.
The general formulas for estimating w and w f are as follows :
Two replications :
E = intrablock error mean square
B = mean square for component 6, based on 2(k 1) degrees
of freedom
w = |, and w' = ^rj
Four replications :
E = intrablock error mean square
B = average mean square of components a and 6, based on
4(fc 1) degrees of freedom
3
Six replications:
E = intrablock error meai square
B = mean square for component a, based on 4(fc 1) degrees
of freedom , Component b need not be
SIMPLE LATTICE EXPERIMENTS 361
, 1
w = -p and
In this problem
19 = i = 381^89 = - 0026186
and
vf =
- E 4(27,747.94) - 3818.89
The weighting factor is
w - w' 0.00023387
w + w f " 0.00028985
= 0.80687
The rfcc x and rfccj, values are multiplied by this weighting factor
to secure the c x > and c v > corrections. Thus
= rk \w + 10') rkCx = 4^
^ J
1 )
y rk \w
In Table 77 are given the average yields of the varieties,
obtained by dividing the totals in Table 74 by 4. The values of
C* and c v > are given at the side and bottom, respectively, of
Table 77. The first c* will be
(0.040344) (-405) = -16.34
The other values of c x / and <v & re calculated in a similar manner.
The adjusted variety means are secured by adding to each
variety average in Table 77 the corrections in the same row and
column. Thus, for variety 1, the adjusted (unconfoimded)
variety mean will be 587.50 - 16.34 + 14.32 = 585.5. Proceed-
ing in a similar manner, the adjusted means of all varieties are
calculated and entered in Table 78.
i As an illustration, consider again the adjustment made for
variety 1. It was shown previously that varieties, 1, 2,, 3, 4, 5
yielded 6260 5855 = 405 g. more in the two blocks containing
this group of varieties than the sum of the yields of the same
varieties grown in different blocks (see Table 73). In like?
362
METHODS OF PLANT BREEDING
manner, varieties 1, 6, 11, 16, 19 yielded 5465 - 5820 -355 g.
less in the two blocks containing them than the sum of the yields
of the same varieties when each was grown in a different block.
TABLE 77. AVERAGE YIELD OF VARIETIES AND c f VALUES
1
587.50
2
572.50
3
586,25
4
660.00
5
622.50
-16.34
6
552 50
7
590.00
8
720.00
9
696 25
10
552.50
-11.10
11
580.00
12
548.75
13
617.50
14
621.25
15
601.25
17.95
16
562.50
17
618.75
18
688.75
19
670.00
20
638.75
-43.37
21
538.75
22
565.00
23
550.00
24
606.25
25
590 00
7 67
c/
14 32
31.06
2.02
-10.29
8.07
TABLE 78. ADJUSTED VARIETY MEANS
1
585.5
2
587.2
3
571.9
4
633.4
5
614.2
6
555.7
7
610.0
8
710.9
9
674.9
10
549.5
11
612.3
12
597.8
13
637.5
14
628.9
15
627.3
16
533.5
17
606.4
18
647.4
19
616.3
20
603.5
21
560.7
22
603.8
23
559.7
24
603.6
25
605.7
The mean of these differences is (405) + (-355) = g g()
tracting 2.50 from the average yield of variety 1, 587.50 (see
Table 77) would give 585.0 as the adjusted mean. This is the
adjusted mean yield given by the original method of analysis
developed by Yates (1936) and illustrated by Yates (1936) and
SIMPLE LATTICE EXPERIMENTS 363
Goulden (1937, 1939). In the present analysis, the interblock
information has been recovered, and the correction must be
multiplied by the weighting factor (w w')/(w + w f ). Sub-
tracting (2.50) (0.80687) from the average yield of 587.50 gives
585.5 as the adjusted mean. The method of computation used in
the problem simplifies the calculation of these correction terms.
To calculate the standard error of the difference between
variety means, the intrablock error mean square (Table 75) is
an estimate of the uncontrolled error variance ($ 2 ) of a single
plot.
The standard error of the difference between the means of two
varieties that have occurred in the same block, such as 1 and 2,
2 and 7, etc., is
1(2) (3818.89) r (2) (.00026186) ]
\ (4) (5) [(.00026186) + (.00002799) """ ( } \
= \/2217.58 = 47.1
The standard error of the difference between the means of two
varieties that did not occur in the same block, such as 1 and 7, is
...
(
The mean standard error of all comparisons is
7 + (k - 1) = V2423J30 = 49.2
Usually, this latter standard error may be used for all compari-
sons of differences between variety means.
If the data had been analyzed as a randomized complete block
(Table 76) the variance of the difference between two variety
means would have been
= 4568 . 23
Assuming the precision obtainable in a randomized-complete-
block design to be 100 per cent, the lattice design would be
364 METHODS OF PLANT BREEDING
4568.23/2423.00 = 189 per cent. This represents a gain of
89 per cent in precision through the use of incomplete blocks.
To make an exact test of significance appropriate to the analy-
sis of a lattice experiment, it is necessary to correct the variety
mean square so that this can be compared with intrablock
error. This requires that the variety mean square be freed of
block differences. To do so, we must first calculate the unad-
justed sum of squares for component b of the blocks. This is
calculated from the totals of the two sets of blocks given in
Table 73. Numerically, this is
(6260) 2 + ' " + (5605) 2 + (546G) 2 + + (5910) 2
10
_ (30,835) 2 + (29,715) 2
50
= 222,136
for 8 degrees of freedom. This may be designated B u . The
adjusted sum of squares for component b (B a ) was given in
Table 75 as 97,705. The adjusted sum of square for varieties
(eliminating blocks) will be the sum of squares for varieties in
Table 75, which is 240,663, minus ( -^ B u - W 7 W \ B\
\ w w + w /
This becomes
240 663 - [(^23387) (222 136) _ 000023387) 1
^u,Qo5 |^ .00026186 ^^ 10U ' .00028985 ^">' UO 'J
= 121,106
Dividing this sum of squares by 24 gives 5046.08 as the adjusted
mean square for varieties. The exact value of F is then 5046. 08/
3818.89 = 1.32,
Lattice experiments can be analyzed as randomized complete
blocks, although when many treatments are included the error
may be rather large. If it is found that B is equal to or less
than E, there would be no advantage in adjusting the variety
means. The unadjusted variety means should be used and the
error obtained from a randomized-complete-block analysis.
When a complete replication is lost the data may be analyzed
as an ordinary randomized-block test. Methods of analysis
appropriate for a lattice design may be used for the yield data,
TRIPLE LATTICE EXPERIMENTS ' 365
the adjusted means being used, and other characters of less
interest may be analyzed as ordinary randomized blocks, the
unadjusted means being used.
TRIPLE LATTICE EXPERIMENTS
In triple lattice experiments, 1 the number of groups is three.
The third group (Z) is added to the X and Y groups used in a
simple lattice design. The number of replications must be a
multiple of three. The number of varieties tested is the square
of some number.
It is always possible to superimpose a Latin-square arrange-
ment on a square of variety numbers. Using the five letters A,
Bj Cj D, E, a Latin square of these letters is superimposed on the
k 2 = 25 varieties, as given below:
IA 2E 3D 4(7 5/?
6# 7A 8E W IOC
11(7 12B 13A UE 151)
16D 17(7 18B 19A 20E
21E 22D 23(7 24B 25A
These 25 varieties may now be arranged in incomplete blocks of
k = 5 varieties each. In the first group, designated X, the five
rows of the square are arranged in random order, and the five
varieties within each row (block) are randomized. The same pro-
cedure with the columns of the square produces the Y group.
For the third, or Z group, the Latin letters are arranged in random
order and the varieties with the same Latin letter randomized
within each Latin-letter block. The blocks in group Z were made
up of varieties with the Latin letters in the order A, E, C, B, D.
A random arrangement in three such groups is given in Table
79.
This random arrangement of "varieties" was superimposed on
the uniformity trial data with rod rows of wheat given by Wiebe
(1935). It was assumed that three row plots per variety were
grown and only the central row harvested. The yields are given
in grams per rod row. Three complete replications were used.
Assembling the yields for the X, F, and Z groups according to
rows, columns, and Latin letters gives Table 80. The block
1 The triple lattice design has been called a two-dimensional pseudofac-
torial arrangement in three groups of sets and a two-dimensional quasi-*
factorial design in randomized blocks in three equal groups of sets,
366
METHODS OF PLANT BREEDING
totals are given also, as well as the total for each complete
replication.
TABLE 79. RANDOM ARRANGEMENT OF VARIETIES IN TRIPLE LATTICE
EXPERIMENT
Replication 1 (Group X) Replication 2 (Group F)
Block Block
(1)
10
7
6
8
9
(2)
14
13
11
15
12
(3)
2
4
5
3
1
(4)
25
24
23
21
22
(5)
18
16
17
20
19
(6)
15
5
10
20
25
(7)
16
6
21
11
1
(8)
2
17
7
22
12
(9)
23
3
13
18
8
(10)
24
4
14
19
9
Replication 3 (Group Z)
Block
(11)
7
1
19
25
13
(12)
14
21
8
2
20
(13)
11
10
23
17
4
(14)
5
24
18
12
6
05)
15
16
9
3
22
If six or nine replications were to be used, the X, Y, and Z
groups would be repeated once or twice, respectively. In
Table 81 is given the sum of the yields of each variety in a
5 by 5 table with appropriate row and column totals. To the
Block
TABLE 80. YIELDS OF VARIETIES IN GRAMS PER ROD Row
Replication 1 (Group X)
Block totals
1
2
3
4
5
(3)
635
525
555
650
635
3,000
6
7
8
9
10
(1)
495
730
810
775
710
3,520
11
12
13
14
15
(2)
630
600
645
635
645
3,155
16
17
18
19
20
(5)
735
690
840
855
805
3,925
21
22
23
24
25
(4)
620
795
590
660
615
3,280
16,880
TRIPLE LATTICE EXPERIMENTS
367
TABLB 80. YIELDS OF VARIETIES IN GRAMS PER ROD Row, (Continued)
Replication 2 (Group Y)
Block Block totals
1
6
11
16
21
(7)
530
490
595
495
540
2,650
2
7
12
17
22
(8)
610
660
620
695
570
3,155
3
8
13
18
23
(9)
705
850
675
685
640
3,555
4
9
14
19
24
(10)
840
905
785
860
875
4,265
5
10
15
20
25
(6)
670
455
655
665
615
3,060
16,685
Block
Replication 3 (Group Z)
Block totals
1
7
13
19
25
(11)
580
675
545
470
570
2,840
6
12
18
24
5
(14)
700
620
575
570
505
2,970
11
17
23
4
10
(13)
515
450
550
495
550
2,560
16
22
3
9
15
(15)
640
700
635
645
640
3,260
21
2
8
14
20
(12)
445
515
465
445
455
2,325
13,955
right of the table are given also the total yields of five varieties
according to the Latin letters.
The analysis of variance may now be calculated. This pro-
cedure will follow closely the form given by Cox, Eckhardt, and
Cochran (1940).
The correction term is (47,520) 2 -s- 75 j 30,108,672. Total
sum of squares is calculated from the sum of the squares of
368
METHODS OF PLANT BREEDING
the 75 individual plot yields minus the correction term. Thus,
31,090,500 - 30,108,672 = 981,828.
TABLE 81. TOTAL YIELD OF VARIETIES
Row Latin
totals letters Totals
1
2
3
4
5
1745
1650
1895
1985
1810
9,085
A
9,660
6
7
8
9
10
1685
2065
2125
2325
1735
9,915
B
9,540
11
12
13
14
' 15
1740
1840
1865
1865
1940
9,250
C
9,055
16
17
18
19
20
1870
1835
2100
2185
1925
9,915
D
10,095
21
22
23
24
25
1605
2065
1780
2105
1800
9,355
E
9,170
Column
totals, .
8645
9455
9765
10,465
9190
47,520
47,520
The sum of squares for replications is
(16,880) 2 + (16,685) 2 + (13,955) 2
25
- 30,108,672 = 213,954
The sum of squares for varieties (ignoring blocks) is calculated
from the variety totals in Table 81. This is
(1745) 2 + (1650) 2 +
3
+ (1800) 2
- 30,108,672 = 260,561
In the triple lattice with three replications, there will be no
component a for blocks, as described for four replications in
the simple lattice design. Component b consists of three sets
of values that may be used to give an estimate of block differ-
ences freed of varietal effects. The first row total in group
Z(3000) contains the yields of varieties 1, 2, 3, 4, 5. An uncon-
founded estimate of the sum of the yield of these five varieties
can be obtained from the first column (Table 80) in group
F(3355) and for these same varieties in group Z(2730). An
estimate of the block effect freed of varietal differences will be
TRIPLE LATTICE EXPERIMENTS 369
given by
2(3000) - 3355 - 2730 = -85
This value can be calculated more conveniently by subtracting
the total of the first row in Table 81 from three times the total
of the first row in Table 80. Thus
3(3000) - 9085 = -85
These values are to be used in making the adjustments to the
variety means. Since it will be easier to add than to subtract,
in making the adjustments, the negative values are determined.
These are designated as 2 rkc x} 2 rkc yj 2 rkc z . They are calculated
as follows:
2 rkc x = row total of Table 81 3 (row total of group X, Table
80)
2rkc y = column total of Table 81 3 (row total of group Y,
Table 80)
2 rkc z = Latin-letter total of table 81 3 (row total of group Z,
Table 80)
The values of 2 rkc are calculated below:
2 rkc g
2 rkcy
9085
- 3(3000)
85
8,645
- 3(2650) =
695
9915
- 3(3520)
= - 645
9,455
- 3(3155) =
- 10
9250
- 3(3155)
= - 215
9,765
- 3(3555) =
- 900
9915
- 3(3925)
= -1860
10,465
- 3(4265) =
-2330
9355
- 3(3280)
= _ 485
9,190
- 3(3060) =
10
-3120
-2535
9,660 - 3(2840) = 1140
9,540 - 3(2970) = 630
9,055 - 3(2560) - 1375
10,095 - 3(3260) = 315
9,170 - 3(2325) = 2195
5655
The sum of the 2 rkc values must be 0.
(-3120) + (-2535) + (5655) =
370
METHODS OF PLANT BREEDING
The sum of squares of the deviations of the 2 rkc values within
sets will be
(85)2 + + (-485)' + (695) 2 + + (10) 2
+ (1140) 2 + + (2195) 2
30
_ (-8120)' + (-2635)' + (6656)' =
J.OU
= 325,429
for 12 degrees of freedom. The divisors are 2 rk and 2 rfc 2 .
The results are now summarized in Table 82.
TABLE 82. ANALYSIS OF VARIANCE OF TRIPLE LATTICE EXPERIMENT
Variation due to
Degrees of
freedom
Sum of
squares
Mean square
Replications
2
213,954
106 977 00
Component b for blocks (eliminat-
ing varieties)
12
325 429
27 119 08
Varieties (ignoring blocks) ...
24
260,561
10,856 71
Error (intrablock)
36
181 , 884
5,052 33
Total
74
981 , 828
A test of significance of variety mean square cannot be made
from the mean squares for varieties and error of the foregoing
analysis, since variety mean square contains some block effects
and an exact test requires additional computation and will be
given later. An approximate test may be made in the form of a
randomized-complete-block analysis by combining the degrees
of freedom and sums of squares for blocks and error in Table 82
to produce Table 83.
TABLE 83. ANALYSIS OF VARIANCE AS RANDOMIZED COMPLETE BLOCKS
Variation due to
Degrees of
freedom
Sum of
squares
Mean
square
F
Replications
2
213,954
106,977 00
Varieties
24
260,561
10,856 71
1 03
Error
48
507,313
10,569.02
Total
74
981,828
TRIPLE LATTICE EXPERIMENTS 371
F = 1.03 is nonsignificant for HI = 24 and n 2 = 48 degrees
of freedom, the varieties being hypothetical. If this test of
significance showed a significant variety mean square, no further
test would be necessary. If, however, the F value did not reach
significance and block mean square in Table 82 were above
error mean square, it would be well to make the exact test of
significance.
To calculate the adjusted variety means the values of 2rkc x ,
2rkc y , and 2rkc z must be multiplied by a weighting factor to
obtain the three sets of corrections <v, ty, arid c z ', respectively.
The weighting factor is ~ ~, where w = 1/E and w' =
2/(3B E), E and B being the error and block mean squares,
respectively, obtained from Table 82.
The general formulas for estimating w and w' are as follows:
Three replications :
E = intrablock error mean square
B = mean square for component 6, based on 3(fc 1) degrees
of freedom Component a does not exisv
1 , , 2
w jz and w = r-^ ~
JLJ O-LJ Ju
Six replications;
E = intrablock error mean square
B = average mean square for components a and b for 6 (k 1)
degrees of freedom
1 , , 5
w = 77 and w = ^-^ r ,
E GB E
Nine replications :
E intrablock error mean square
B = mean square for component a for G(fc 1) degrees of
freedom. Component b need not be used
w = -=j and w f = -^
If B is less than or equal to E, the randomized-complete-block
analysis is to be used. The unweighted variety means are then
tested with the error calculated from the randomized-complete-
alock analysis.
In this problem, the mean square for blocks (Table 82) is much
greater than error mean square, and the weighting of the variety
372
METHODS OF PLANT BREEDING
means will lead to an increase in precision. Referring to the
mean squares in Table 82,
w = _. . =
1
w =
E 5052.33
2
= 0.00019793
2
W - E 3(27,119.08) - 5,052.33
The weighting factor is then
' = 0.81370
= 0.00002621
w
The weighted correction terms will be
w
= 0.027123(2rtc.)
<v =
1 2 <>J^) (2rkc v ) = 0.027l23(2rkc y )
2rk 2w + w f v y/ x
1 2(ti? w 1
2r/c
(2rA?c,) = 0.
In Table 84 are given the average yields of the three replica-
tions of each variety, found by dividing the total yields in
TABLE 84. AVERAGE YIELDS AND c f VALUES
1A
581 . 67
2E
550.00
3D
631 67
4C
661 . 67
5B
603.33
2.31
6B
561.67
7A
688.33
8E
708.33
9D
775.00
IOC
571.67
-17.49
11C
580 00
12B
613.33
13 A
621.67
14E
621 . 67
15D
646.67
- 5.83
16D
623.33
17C
611.67
18B
700.00
19A
728.33
20E
641 . 67
-50.45
21E
535.00
22D
688.33
23C
593.33
24B
701.67
25A
600.00
-13.15
<y
18.85
-0.27
-24.41
-63.20
0.27
<Y
A
30.92
B
17.09
C
37.29
D
8.54
E
59,53
TRIPLE LATTICE EXPERIMENTS
373
Table 81 by 3. The weighted corrections c x >, <v, and c z > are
obtained by multiplying the values of 2rkc XJ 2rkc y and 2rkc z by
0.027123. The first <v is
(0.027123) (85) = 2.31
These are recorded for the proper row, column, or Latin letter in
Table 84.
The adjusted mean yields are obtained by adding to the aver-
age yield of each variety the correction term in the same row,
column and for the same Latin letter. For variety 1, the
adjusted mean is 581.67 + 2.31 + 18.85 + 30.92 = 633.8. Pro-
ceeding in a similar manner for all varieties gives the adjusted
means in Table 85.
TABLE 85. ADJUSTED VAKIETY MEANS
1
633.8
2
611.6
3
618.1
4
638.1
5
623.0
6
580.1
7
701.5
8
726.0
9
702.9
10
591 7
11
630.3
12
624.3
13
622.4
14
612.2
15
649 7
16
600.3
17
598.2
18
642.2
19
645.6
20
651
21
600.2
22
683.5
23
593 1
24
642.4
25
618
The mean of the 25 adjusted variety means will be the same as
the mean of the 25 unadjusted means. Three times the total
of Table 85 will be the grand total of Table 81.
Using r 3 replications, k = 5 plots per block, and s 2 =
5052.33 (the error mean square in Table 82), the standard error
of the difference between the adjusted mean yields of two varie-
ties having occured in the same block will be
2w + w'
+ /3L
^/v
'2(5052.33)
(3)(5)
6(.00019793)
(5-
\2(.00019793) + .00062621
= V3916.36 = 62.6
374 METHODS OF PLANT BREEDING
The standard error of the difference between two varieties
that did not occur in the same block is
>/|{HTT? + <*-}-
The mean standard error of the difference of all comparisons is
= 63.7
n l n--7
r(k + 1) \2w + w '
This latter standard error usually may be used for all compari-
sons without appreciable error.
To test the efficiency of the triple lattice design, we may divide
the error variance of the difference between two variety means
as calculated from the randomized-complete-block analysis
by the error variance obtained through use of the triple-lattice
arrangement.
The variance of the difference between two means by random-
. , , . , . . , . 1 , u 2s 2 2(10,569.02)
ized-complete-block design would be = -= =
T o
7046.01. Dividing 7046.01 by 4053.41 gives 174 per cent as the
precision of the lattice design, when the ordinary randomized
block is considered 100 per cent. Reducing the block size from
25 plots per complete block to 5 plots per incomplete block
resulted in a gain in precision of 74 per cent.
The exact test of significance appropriate for the triple lattice
design will now be illustrated. To make this test, it is necessary
to calculate the variety mean square freed of block effects. The
sum of squares for varieties (ignoring blocks), given in Table 82,
must be diminished by
2(w - w') K __ 2(w - w')
2w u "~2w + w r a
where B u and B a are thjs unadjusted and adjusted sums of squares,
respectively, for component b of the blocks.
B u will be the sum of squares between blocks, within sets,
calculated from the block totals in Table 80. Thus
(3000) 2 + + (3280) 2 + (2650) 2 + ' ' ' + (3060) 2
+ (2840) 2 + - + (2325) 2
5
+ (36,685) 2
25
DATA EXPRESSED AS PERCENTAGES 375
B a was given in Table 82 as 325,429. Then
2^0 ^ _ ^-) ^ = (0 . 8675?9)(507)404)
- (0.813704) (325,429) = 175,410
Subtracting this quantity from the unadjusted sum of squares
for varieties gives 260,561 - 175,410 = 85,151 as the adjusted
sum of squares for varieties. Dividing by 24 degrees of freedom
gives 3547.96 as the corrected mean square. Since this is less
than the mean square for error (5052.33, Table 82), F will be less
than 1. A lack of significance is clearly indicated.
ANALYSIS OF DATA EXPRESSED AS PERCENTAGES
In the analysis of data expressed in the form of a binomial,
such as is frequently encountered in studies of the proportion, or
percentage, of plants diseased, the standard error of the pro-
portion will be given by \/pq/N, where p is the proportion
diseased, q = (1 p) is the proportion disease free, and N is the
total number of plants in the sample. If, for example, one-fourth
of the plants in a sample of 1 00 were diseased, the standard error
of p = 0.25 would be = V0.001875 = 0.043. If
expressed as p = 25 per cent, the standard error would be
4.3 per cent.
The standard error of a proportion, or percentage, is clearly
dependent on the value of p as well as N, being a maximum when
p, 0.50 and reducing to zero as p becomes or 1.00. Since a
basic assumption in the use of a generalized error from an analysis
of variance is that the errors of the separate treatments must be
independent of the means, data expressed as percentages fre-
quently need to be transformed before being analyzed by means
of an analysis of variance. The transformation would need to
be one for which the variance of each treatment is equalized and
dependent on N alone. Bliss (1937) suggested an angular
transformation in which p is replaced by sin 2 0, Tables for
making such transformations have been given by Bliss (1937,
1938), Fisher and Yates (1938), and Snedecor (1940) and are
reproduced as Appendix Table VI. A discussion of some of the
difficulties in the analysis of data of this type has been given by
Cochran (1938).
376
METHODS OF PLANT BREEDING
The use of such transformations will be illustrated with data
taken from a paper by Salmon (1938). Clark and Leonard
(1939) have given a full analysis of these data. In Table 86 is
given the percentage of bunt on each of 5 varieties of wheat
inoculated with 10 different collections of the organism. Two
replications were used. The percentages were based on counts
of 200 to 400 heads per plot.
TABLE 86. PERCENTAGE INFECTION IN DIFFERENT VARIETIES OF WHEAT
WITH 10 COLLECTIONS OF BUNT, IN EACH OF Two REPLICATIONS
(FROM SALMON)
Bunt
Hybrid 128
Mmturki
Turkey
Albit
Ridit
col-
lec-
Total
tion
num-
I
II
I
II
I
II
I
II
I
II
ber
2
76
95
91
84
89
84
92
91
9
2
713
3
95
93
88
75
3
6
94
90
6
1
551
4
91
92
92
83
82
87
14
5
4
3
553
5
84
90
61
81
8
2
4
3
4
4
341
7
98
98
56
44
14
9
1
2
2
324
10
94
83
71
64
6
1
92
80
4
4
499
11
83
78
71
70
4
1
2
4
7
6
326
51
94
96
45
40
28
22
1
3
5
5
339
157
75
86
. 75
85
52
92
89
85
1
1
641
189
87
95
81
80
80
92
92
95
5
6
713
Total
877
906
731
706
366
396
480
457
47
34
5000
These data were transformed into the form p = sin 2 6 by
means of Appendix Table VI. The transformed data are given
in Table 87.
These transformed data in Table 87 may be subjected to an
(4361. 7) 2
analysis of variance. The correction term will be
100
190,244.27. The total sum of squares will be 269,490.93 -
190,244.27 = 79,246.66. The sum of squares for replication is
obtained by adding the totals for replicates 1 and 2, respectively,
for all five varieties to give 2188.4 and 2173.3. The sum of
f r .. . (2188.4) 2 + (2173.3) 2 . .,
squares for replication is ^ "~~^ minus the cor-
50
rection term or 2,28.
DATA EXPRESSED AS PERCENTAGES
377
In Table 88 are added the values of the transformations for the
two replications to give the totals for both.
TABLE 87. PERCENTAGE DATA FROM TABLE 86 TRANSFORMED TO DEGREES
BY MEANS OP THE TRANSFORMATION p = siN 2 6
Bunt
col-
Hybrid 128
Minturki
Turkey
Albit
Ridit
lec-
Total
tion
num-
I
II
I
II
I
II
I
II
I
II
ber
2
60 7
77 1
72 5
66.4
70.6
66 4
73 6
72.5
17 5
8 1
585.4
3
77.1
74 7
69 7
60
10
14 2
75 8
71 6
14 2
5 7
473
4
72.5
73.6
73 6
65.7
64 9
68 9
22
12.9
11.5
10.0
475 6
5
66.4
71.6
51 4
64.2
16.4
8.1
11.5
10.0
11.5
11 5
322.6
7
81.9
81 9
48.5
41.6
22
17 5
0.0
5 7
8.1
8 1
315 3
10
75.8
65.7
57.4
53.1
14.2
5.7
73.6
63.4
11.5
11.5
431.9
11
65.7
62.0
57 4
56.8
11.5
5 7
8.1
11.5
15.3
14 2
308 2
51
75 8
78.5
42 1
39 2
32
28
5.7
10
12.9
12.9
337.1
157
60.0
68
60.0
67 2
46.2
73.6
70.6
67.2
5.7
5 7
524 2
189
68 9
77.1
64 2
63.4
63.4
73 6
73 6
77.1
12 9
14.2
588.4
Total
704 8
730 2
596.8
577.6
351 .2
361 7
414 5
ibT~9
121.1
101 9
4361 7
I
TABLE 88. TOTALS OF Two REPLICATIONS FOR EACH VARIETY AND BUNT
COLLECTION
Bunt col-
lection
Hybrid
128
Minturki
Turkey
Albit
Ridit
Total
number
2
137 8
138 9
137
146 1
25.6
585.4
3
151.8
129 7
24 2
147 4
19 9
473.0
4
146 1
139 3
133 8
34 9
21 5
475 . 6
5
138.0
115.6
24 5
21 5
23.0
322.6
7
163.8
90 1
39 5
5 7
16 2
315.3
10
141.5
110 5
19 9
137.0
23.0
431.9
11
127.7
114.2
17.2
19 6
29.5
308.2
51
154 3
81 3
60
15 7
25 8
337.1
157
128
127.2
119 8
137.8
11 .4
524.2
189
146.0
127 6
137
150 7
27.1
588.4
Total. . .
1435.0
1174.4
712.9
816.4
223.0
4361 . 7
The sum of squares for bunt collections will be
(58S.4) 2 + + (5S8.4) 2
10
- 190,244.27 = 10,982.11
378
METHODS OF PLANT BREEDING
The sum of squares for varieties will be
(1435.0) 2 + + (223.0) 2
20
- 190,244.27 = 42,900.97
The sum of squares for varieties inoculated with the individual
bunt collections is calculated from the 50 figures within Table 88.
Thus (137.8)' + (138.9)' + ....+(27.11' _ 190 ,244.27 =
77,937.02. The sum of squares for interaction of varieties X
collections will be
77,937.02 - 10,982.11 - 42,900.97 = 24,053.94
The analysis of variance is given in Table 89.
TABLE 89. ANALYSIS OF VARIANCE OP TRANSFORMED DATA
Variation due to
Degrees of
freedom
Sum of
squares
Mean
square
F
Blocks
1
2 28
2 28
Varieties
4
42,900 97
10,725 24
402 00*
Collections, ,
9
10,982.11
1,220 23
45 74*
Varieties X collection . . .
Krror
36
49
24,053 94
1,307.36
668.17
26 68
25.04*
Total
99
79,246 66
* Exceeds the 1 per cent point.
The varieties gave highly significant differences in their average
reaction to all collections of bunt. The bunt collections differed
significantly in their ability to produce the disease, as an average of
all varieties. Furthermore, the interaction of varieties X collec-
tions was highly significant, indicating clearly that these collec-
tions produced differential responses on the different varieties.
The standard error of the difference between the means of varie-
ties for all bunt collections would be V(2 X 26768) /20 = 1.63.
The totals in table could be compared also. The standard error
of the difference between means is \/2s 2 /N, and the standard error
of the difference between totals is -\/2s 2 N. The standard
error of the difference between two variety totals would be
A/2 X 26.68 X 20 = 32.67.
The standard error for determining significant interactions
of the totals for two replications would be \/2 X 2 X s 2 X N =
DATA EXPRESSED AS PERCENTAGES 379
V2 X 2 X 26.68 X 2 = 12.65. The difference between the dif-
ferences in reaction of Turkey and Albit to collections 3 and 4 is
seen to be (24.2 - 133.8) - (147.4 - 34.9) - -221.1. Since
this difference is 15.1 times its standard error, it is clear that
these two varieties reacted in a differential manner to collections
3 and 4. Other comparisons could be made in a similar manner.
If the range in the percentages is between about 25 and 75, no
transformation probably would need to be made. In this range,
the errors of the separate varieties would be sufficiently similar
so that a transformation would be unnecessary. If, however,
the range in percentages goes below 25 or above 75, a transforma-
tion probably would be worth while. Such would be true par-
ticularly if some of the percentages were very low or very high.
LITERATURE CITATIONS
AAMODT, 0. 8., 1923. The inheritance of growth habit and resistance to
stem rust in a cross between two varieties of common wheat. Jour.
Agr. Research, 24: 457-469.
AMERICAN ASSOCIATION OF CEREAL CHEMISTS, 1941, Cereal Laboratory
Methods, Ed. 4. Lund Press, Minneapolis.
ANDERSON, D. C., 1938. The relation between single and double cross
yields in corn. Jour. Am. Soc. Agron., 30: 209-211.
ARNY, A. C., 1921 . Further experiments in field technics in plot tests. Jour.
Agr. Research, 21: 483-499.
, 1922. Border effect and ways to avoid it. Jour. Am. Soc. Agron.j
14: 266-278.
, 1936. Breeding better flax varieties for Minnesota. Proc. Minn,
Acad. Sri., pp. 29-38.
-, and H. K. HAYES, 1918. Experiments in field technic in plot tests.
Jour. Agr. Research, 16: 251-262.
ABHBY, ERIC, 1930. Studies on the inheritance of physiological characters.
I. A physiological investigation of the nature of hybrid vigour in maize.
Ann. Bot.j 44: 457-467.
, 1932. Studies in the inheritance of physiological characters. II.
Further experiments upon the basis of hybrid vigour and upon the
inheritance of efficiency index and respiration rate in maize. Ann. Bot.,
46: 1007-1032.
, 1937. Studies in the inheritance of physiological characters. III.
Hybrid vigour in the tomato. Part 1 . Manifestations of hybrid
vigour from germination to the onset of flowering. Ann. Bot. (N.S.)>
1:11-41.
ATKINS, I. M., 1938, Relation of certain plant characters to strength of
straw and lodging in winter wheat. Jour. Agr. Research, 56: 99-120.
ATWOOD, S. S., 1940. Genetics of cross-incompatibility among self-incom-
patible plants of Trifolium repens. Jour. Am. Soc. Agron., 32 : 955-968.
, 1941. Cytogenetic basis of self-compatibility in Trifolium repens.
Genetics, 26: 137.
ATJSEMUS, E. R., 1934. Correlated inheritance of reaction to diseases and
of botanical characters in triangular wheat crosses. Jour. A gr. Research,
48: 31-57.
, M. C. MARKLEY, C. H. BAILEY, and H. K. HAYES, 1938. Quality
studies in the wheat breeding program at the Minnesota Agricultural
Experiment Station. Jour. Agr. Research, 56: 453-464.
BARKER, H. D., 1923. A study of wilt resistance in flax, Minn. Agr. Exp.
Sta. Tech. Bull. 20.
381
382 METHODS OF PLANT BREEDING
BARTLETT, M. S., 1937. Some examples of statistical methods of research
in agriculture and applied biology. SuppL Jour. Royal Stat. Soc., 4:
J 37-1 83.
BAUR, E., 1931. Masters Memorial Lectures. I. Evolution. II. Scope
and methods of plant-breeding work at the Kaiser Wilhelm Institute
fur Ziichtungsforshung, Muncheberg (Mark). Jour. Royal Hort. Soc.,
66: 176-190.
, 1933. Die heutige Stand der Rebenzuchtung in Deutschland. Der
Zuchter, 5 : 73-77.
BEDDOWS, A. R., 1931. Seed setting and flowering in various grasses.
Welsh Plant Breeding Station Bull., Series H, No. 12, pp. 5-99.
BIFFIN, R. H., 1905. Mendel's laws of inheritance and wheat breeding.
Jour. Agr. Sri., 1: 4-48.
, 1916. The suppression of characters on crossing. Jour. Genetics,
5:225-228.
BINDLOSS, ELIZABETH A., 1938. Nuclear size in plumular meristems of
inbred and hybrid maize. Am. Jour. Bot., 25: 738-743.
BLAKESLEE, A. F., 1939. The present and potential service of chemistry to
plant breeding. Am. Jour. Bot., 26: 163-172.
BLANCHARD, R. A., J. H. BIGGER, and R. O. SWELLING, 1941. The resist-
ance of corn strains to the corn ear worm. Jour. Am. Soc. Agron., 33:
344-350.
BLISS, C. I., 1937. The analysis of field experimental data expressed in
percentages. Plant Protection Bull. 12, pp. 67-77. U.S.S.R. (In
Russian, with English summary.)
, 1938. The transformation of percentages for use in the analysis
of variance. Ohio Jour. Sci., 38: 9-12.
BLODGETT, F. M., and KARL FERNOW, 1921. Testing seed potatoes for
mosaic and leaf roll. Phytopathology, 11 : 58-59.
BOLLEY, H. L., 1901. Flax and flax seed selection. North Dak. Agr. Exp,
Sta. Bull 55.
, 1912. Resistant flax seed for sowing purposes. North Dak. Agr.
Exp. Sta. Press Bull. 57.
BORGESON, CARL, and H. K. HAYES, 1941. The Minnesota method of seed
increase and seed registration for hybrid corn. Jour. Am. Soc. Agron.,
33:70-74.
BREWBAKER, H. E., 1926. Studies of self-fertilization in rye. Minn. Agr.
Exp. Sta. Tech. Bull. 40.
, and F. R. IMMER, 1931. Variations in stand as sources of experi-
mental error in field tests with corn. Jour. Am. Soc. Agron., 23 : 469-
480.
BRIEGER, FRIEDRICH, 1930. Selbststerilitat und Kreuzungssterilitat in
Pflanzenreich und Tierreich. Julius Springer, Berlin.
BRIGGS, F. N., 1930. Breeding wheats resistant to bunt by the backcross
method. Jour. Am. Soc. Agron., 22: 239-244.
, 1933, A third genetic factor for resistance to bunt, Tilletia tritici,
in wheat hybrids. Jour. Genetics, 27: 435-441.
LITERATURE CITATIONS 383
, 1 934. Inheritance of resistance to bunt, Tilletia tritici, in Sherman
and Oro wheat hybrids. Genetics, 19 : 73-82.
, 1935. The backcross method in plant breeding. Jour. Am. Soc.
Agron., 27 : 971-973.
, 1938. The use of the backcross in crop improvement. Am.
Naturalist, 72 : 285-292.
1940. Linkage between the Martin and Turkey factors for resist-
ance to bunt, Tilletia tritici, in wheat. Jour. Am. Soc. Agron., 32 : 539-
541.
BRINK, R. A., and D. C. COOPER, 1939. Soirmtoplastic sterility in Medicago
saliva. Science, 90: 545-546.
BROADFOOT, W. C., 1926. Studies on the parasitism of Fusarium lini
Bolley. Phytopathology, 16: 951-978.
BROOKINS, W. W., 1940. Linkage relations of the factors determining reac-
tion to stem rust in barley. Ph.D. thesis. University of Minnesota.
BRYAN, A. A., 1933. Factors affecting experimental error in field plot tests
with corn. Iowa Agr. Exp. Sta., Bull. 163.
BTJRNHAM, C. R., 1932. The inheritance of Fusarium wilt resistance in flax.
Jour. Am. Soc. Agron., 24: 734-748.
, and J. L. CARTLEDE, 1939. Linkage relations between smut resist-
ance and semi-sterility in maize. Jour. Am. Soc. Agron., 31: 924933.
BUSHNELL, J. W., 1922. Isolation of uniform typos of Hubbard squash by
inbreeding. Proc. Amer. Soc. Hort. ScL, 19: 139-144.
CAMPBELL, H. A., W. L. ROBERTS, W. K. SMITH, and K. P. LINK, 1940.
Studies on the hemorrhagic sweet clover disease. I. The preparation
of hemorrhagic concentrates. Jour. Biol. Chem., 136: 47-55.
CHANG, S. C., 1940. Morphological and physiological causes for varietal
differences in shattering and after harvest sprouting in cereal crops.
Ph.D. thesis. University of Minnesota.
CHRISTIDIS, B. G., 1931. The importance of the shape of plots in field
experiments. Jour. Agr. ScL, 21: 14-37.
CHURCHWARD, J. G., 1931. Studies in the inheritance of resistance to bunt
in a cross between Florence and Hard Federation wheats. Proc. Hoy.
Soc. N.S.W., 64: 298-319.
, 1932. Inheritance of resistance to bunt, Tilletia tritici (Bjerk).
Winter, and other characters in certain crosses of Florence wheat.
Proc. Linn. Soc. N.S.W., 57: 133-147.
CLARK, ANDREW, and W. H. LEONARD, 1939. The analysis of variance
with special reference to data expressed as percentages. Jour. Am.
Soc. Agron., 31: 55-66.
CLARK, E. R., and H. K. WILSON, 1933. Lodging in small grains. Jour.
Am. Soc. Agron., 25: 561-572.
CLARK, J. A., 1936. Improvement in wheat. U. S. Dept. Agr. Yearbook,
pp. 207-302.
, and E. R. AUSEMUS, 1928. Immunity of Hope wheat from black
stem rust inherited as a dominant character. Jour. Am. Soc. Agron..
20: 152-159.
384 METHODS OF PLANT BREEDING
CLARKE, S. E., 1927. Self-fertilization in timothy. Sci. Agr., 7: 409-439.
CLAYTON, J. S., and R. K. LARMOUR, 1935. A comparative color test for
coumarin and melilotic acid in Melilotus species. Can. Jour. Research,
13 : 8&-10Q.
COCHRAN, W. G., 1937. A catalogue of uniformity trial data. Suppl. Jour.
Roy. Stat. Soc., 4 : 233-253.
, 1938. Some difficulties in the statistical analysis of replicated
experiments. Emp. Jour. Exp. Agr., 6: 157-175.
COFFMAN. F. A., J. H. PARKER, and K. S. QUISENBERRY, 1925. A study of
variability in the Hurt oat. Jour. Agr. Research, 30: 1-64.
COLLINS. G, N., 1921. Dominance and the vigor of first generation hybrids.
Am. Naturalist, 65: 116-133.
, and J. H. KEMPTON, 1917. Breeding sweet corn resistant to the
corn ear worm. Jour. Agr. Research, 11: 549-572.
COLLINS. J. L., arid K. R. KERNS, 1938. Mutations in the pineapple. Jour.
Heredity, 29: 163-172.
COOPER, D. C., arid R. A. BRINK, 1940. Partial self-incompatibility and the
collapse of fertile ovules as factors affecting seed formation in alfalfa.
Jour. Agr. Research, 60: 453-472.
COOPER, H. P., 1923. The inheritance of spring and winter habit in crosses
between typical spring and typical winter wheats and the response of
wheat to artificial lights. Jour. Am. Soc. Agron,, 16: 15-25.
Cox, GERTRUDE M., R. C. ECKHARDT, and W. G. COCHRAN, 1940. The
analysis of lattice and triple lattice experiments in corn variety trials.
Iowa Agr. Exp. Sta. Res. Bull. 281.
CRAIGIE, J. H., 1940. The origin of physiologic races of rust fungi through
hybridization. The Genetics of Pathogenic Organisms, Publication of
the American Association for the Advancement of Science, No. 12, pp.
66-72. The Science Press.
CRANE, M. B., and W. J. C. LAWRENCE, 1934. The Genetics of Garden
Plants. Macmillan & Company, Ltd., London.
CUMMINGS, M. B., and E. W. JENKINS, 1928. Pure line studies with ten
generations of Hubbard squash. Vermont Agr. Exp. Sta., Bull. 280.
CURTIS, L. C., 1939. Heterosis in summer squash (Cucurbita pepo) and
the possibilities of producing F\ hybrid seed for commercial planting.
Proc. Am. Soc. Hort. Sci., 37: 827-828.
CUTLER, G. H., and W. W. WORZELLA, 1931. A modification of the
Saunders' test for measuring "quality" of wheats for different pur-
poses. Jour. Am. Soc. Agron. , 32: 1000-1009.
, and W. W. WORZELLA, 1933. The wheat-meal fermentation time
test of "quality " in wheat as adapted for small plant breeding samples.
Cereal Chemistry, 10 : 250-262.
DARLINGTON, C. D., 1927. The behavior of polyploids. Nature, 119 : 390-
391.
DARWIN, CHARLES, 1876. The Effects of Cross- and Self-fertilization in the
Vegetable Kingdom. D. Appleton-Century Company, Inc., New York.
DAVIS, R. L., 1929. Report of the plant breeder. Porto Rico Agr. Exp.
Sta., Ann. Rept. 1927: 14-15.
LITERATURE CITATIONS 385
DEEMBN, HAIG, 1940. Colchicine polyploidy and technique. Botanical
Review, 6 : 599-635.
DICKSQN, J. G., 1939. Diseases of Cereal and Forage Crops. Burgess
Publishing Co., Minneapolis, Minn.
DILLMAN, A. C., 1936. Improvement in flax. U. 8. Dept. Agr. Yearbook,
pp. 745-784.
DOXTATOE, C. W., and I. J. JOHNSON, 1936. Prediction of double cross
yields in corn. Jour. Am. Soc. Agron,, 28 : 460-462.
EAST, E. M., 1909. A note concerning inheritance in sweet corn. Science
29:465-467.
, 1929. Self-sterility. BMiog. Gwetica, 5: 331-370.
, 1934. The reaction of the stigmatic tissue against pollen-tube
growth in self -sterile plants. Proc. Nat. Acad. Sci., 20: 364-368.
, 1935a. Genetic reactions in Nicotiana. I. Compatibility.
Genetics, 20: 403-413.
, 19356. Genetic reactions in Nicotiana. IT. Phenotypic reaction
patterns, (rentes, 20: 414-442.
, 1935r. Genetic reactions in Nicotiana. III. Dominance.
Genetics, 20: 443-451.
, 1936a. Genetic aspects of certain problems of evolution. Am.
Naturalist, 70: 143-158.
, 19366. Heterosis. Genetics, 21: 375-397.
, and H. K. HAYES, 1912. Heterozygosis in evolution and in plant
breeding. U.S. Dept. Agr., Bur. Plant Ind. Bull. 243.
, and D. F. Jones, 1919. Inbreeding and Outbreeding. J. B. Lippin-
cott Company, Philadelphia.
-, and D. F. JONES, 1921. Genetic studies on the protein content of
maize. Genetics, 5: 543-610.
ECKHARDT, R. C., and A. A. BRYAN, 1940. Effect of method of combining
the four inbred lines of a double cross of maize upon the yield and
variability of the resulting hybrid. Jour. Am. Soc. Agron., 32 : 347-353.
EDGERTON, C. W., 1918. A study of wilt resistance in the seedbed.
Phytopathology, 8: 5-14.
EMERSON, R. A., G. W. BEADLE, and A. C. FRASER, 1935. A summary of
linkage studies in maize. N.Y. (Cornell) Agr. Exp. Sta., Mem. 180.
, and E. M. EAST, 1913. The inheritance of quantitative characters
in maize. Nebr. Agr. Exp. Sta. Res. Bull. 2.
EMSWELLER, S. L., and PHILIP BRIE RLE Y, 1940. Colchicine-induced
tetraploidy in Lilium. Jour. Heredity, 31 : 223-230.
, and H. A. JONES, 1934. The inheritance of resistance to rust in the
snapdragon. Hilgardia, 8: 197-211.
ENGLEDOW, F. L., 1914. A case of repulsion in wheat. Proc. Cambridge
Phil. Soc., 17:433-435.
, 1920. The inheritance of glume length and grain length in a wheat
cross. Jour. Genetics^ 10: 109-134.
, 1924. Inheritance in barley. III. The awn and the lateral floret
(cont/d): Fluctuation: A linkage: Multiple allelomorphs. Jour.
Genetics, 14 : 49-87.
386 METHODS OF PLANT BREEDING
FISHER, R. A., 1937. The Design of Experiments, Ed. 2. Oliver & Boyd,
Edinburgh.
, 1938. Statistical Methods for Research Workers, Ed. 7. Oliver <fe
Boyd, Edinburgh.
, and F. YATBS, 1938. Statistical Tables for Biological, Agricultural
and Medical Research. Oliver & Boyd, Edinburgh.
FLOR, H. H., 1940. New physiologic races of flax rust. Jour. Agr.
Research, 60: 575-591.
FRASEB, A. C., 1919. The inheritance of weak awn in certain Avena crosses
and its relation to other characters of the oat grain. N.Y. (Cornell)
Agr. Exp. 8ta. Mem,, 23, pp. 635-676.
FREEMAN, G. F., 1918. Producing breadmaking wheats in warm climates.
Jour. Heredity, 9: 211-226.
GATNKS, E. F., 1917. Inheritance in wheat, barley, and oat hybrids.
Wash. Agr. Exp. Sta. Bull 135.
, 1934. The results of recent research in genetics. Proc. 5th Par.
Sci. Cong., pp. 2627-2629. The Toronto Press.
~, and HANNAH C. AASE, 1926. A haploid wheat plant. Am. Jour.
BoL, 13:375-385.
GARBKR, R. J., 1922. Inheritance and yield with particular reference to rust
resistance and panicle type in oats. Minn. Agr. Exp. Sta. Tech. Bull. 7.
, and M. M. HOOVER, 1930. Persistence of soil differences with
respect to productivity. Jour. Am. Soc. Agron., 2: 883-890.
, T. C. MclLVAiNE, and M. M. HOOVER, 1926. A study of soil
heterogeneity in experiment plots. Jour. Agr. Research, 33 : 255-268.
-, and K. S. QUISKNBERRY, 1925. Breeding corn for resistance to
smut (Ustilago zeae). Jour. Am. Soc. Agron., 17: 132-140.
GARNER, W. W., II. A. ALLARD, and E. E. CLAYTON, 1936. Superior germ-
plasm in tobacco. U.S. Dept. Agr. Yearbook, pp. 785-830.
GARRISON, H. S., and F. D. RICHEY, 1925. Effects of continuous selection
for ear type in corn. U.S. Dept. Agr. Bull. 1341.
GOULDEN, C. H., 1931. Modern methods in field experimentation. Sci.
Agr., 11:681-701.
, 1937. Modern methods of testing a large number of varieties.
Dominion of Canada Dept. of Agr. Tech. Bull. 9.
, 1939. Methods of Statistical Analysis. John Wiley & Sons, Inc ,
New York.
, K. W. NEATBY, and J. W. WELSH, 1928. The inheritance of resist-
ance to Puccinia graminis tritici in a cross between two varieties of
Triticum vulgare. Phytopathology, 18 : 631-658.
GRIFFEE, FRED, 1925. Correlated inheritance of botanical characters in
barley, and manner of reaction to Helminthosporium sativum. Jour.
Agr. 'Research, 30: 915-933.
HABBR, E. S., 1929. Inbreeding the table queen (Des Moines) squash.
Proc. Amer. Soc. Hort. ScL, 25: 111-114.
^ ] 938 A study of drought resistance in inbred strains of sweet corn,
Zea mays var. Rugosa. Iowa Agr. Exp. Sta. Res. Bull 243.
LITERATURE CITATIONS 387
HALL, D. M., 1934. The relationship between certain morphological
characters and lodging in corn. Minn. Agr. Exp. Sta. Tech. Bull. 103.
HAMILTON, R. I,, 1926. Improving sunflowers by inbreeding, $a. Agr., 6:
190-192.
HARLAN, H. V., 1918. The identification of varieties of barley. U.S. Dept.
Agr. Bull 622.
, and H. K. HAYES, 1919. Breeding small grains in Minnesota.
II. Barley investigations. Minn. Agr. Exp. Sta. Bull. 182.
, and H. K. HAYES, 1920. Occurrence of the fixed intermedium,
Hordeum intermedium haxtoni, in crosses between //. vulgare pallidum
and II. distichon palmella. Jour. Agr. Research, 19: 575591.
-, and M. L. MARTINI, 1936. Problems and results in barley breeding.
U.8. Dept. Agr. Yearbook, pp. 303-346.
,_ ^ am i M. L. MARTINI, 1940. A study of methods in barley breeding.
U.8. Dept. Agr. Tech. Bull. 720.
and M. N. POPE, 1922. The use and value of backcrosses in small
grain breeding. Jour. Heredity, 13: 319-322.
HARRINGTON, J. B., 1932. Predicting the value of a cross from an F 2
analysis. Canadian Jour. Research, 6: 21-37.
-, 1937. The mass-pedigree method in the hybridization improve-
ment of cereals. Jour. Am. 8oc. Argon., 29: 379-384.
, 1940. Yielding capacity of wheat crosses as indicated by bulk
hybrid tests. Canadian Jour. Research, 18: 578584.
, and O. S. AAMODT, 1923. Mode of inheritance of resistance to
Puccinia graminis with relation to seed color in crosses between varieties
of durum wheat. Jour. Agr. Research, 24: 979-996.
, and P. F. KNOWLES, 1940a. Dormancy in wheat and barley varie-
ties in relation to breeding. Set. Agr., 20: 355-364.
, and P. F. KNOWLES, 19406. The breeding significance of after-
harvest sprouting in wheat. Sci. Agr., 20: 402-413.
HARRIS, J. A., 1 91 2. A simple test of the goodness of fit of Mendelian ratios.
Am. Naturalist, 46: 741-745.
_ , y 1915. On a criterion of substratum homogeneity (or heterogeneity)
in field experiments. Am. Naturalist, 49: 430-454.
, 1920. Practical universality of field heterogeneity as a factor influ-
encing plot yields. Jour. Agr. Research, 19: 279-314.
-, and C. S. SCOFIELD, 1920. Permanence of differences m the plots
of an experimental field. Jour. Agr. Research, 20: 335-356.
, and C. S. SCOFIELD, 1928. Further studies on the permanence of
differences in the plots of an experimental field. Jour. Agr. Research,
36; 15-40.
HARVEY, P. H., 1939. Hereditary variation in plant nutrition. Genetics,
24:437-461.
HAUGE, S. M., and J. F. TROST, 1930. An inheritance study of the distri-
bution of vitamin A in maize. Jour. Biol. Chemistry, 86 : 167-172.
HAYES, H. K., 1922. Production of high protein maize by Mendelian
methods. Genetics, 1 : 237-257.
388 METHODS OF PLANT BREEDING
, 1926. Breeding improved varieties of smooth-awned barleys.
Jour. Heredity, 17: 370-381.
, and O. B. AAMODT, 1927. Inheritance of winter hardiness and
growth habit in crosses of Marquis with Minhardi and Minturki wheats.
Jour. Agr. Research, 36: 223-236.
, and C. L. ALEXANDER, 1924. Methods of corn breeding. Minn.
Agr. Exp. Sta. Bull. 210.
, and A. C. ARNY, 1917. Experiments in field technics in rod row
tests. Jour. Agr. Research, 11: 399-419.
, E. R. AUSEMUS, E. C. STAKMAN, C. H. BAILEY, H, K. WILSON,
R. H. BAMBERO, M. C. MARKLEY, R. F. CRIM, and M. N. LEVINE,
1936. Thatcher wheat. Minn. Agr. Exp. Sta. Bull. 325.
, and E. M. EAST, 1915. Further experiments on inheritance in
maize. Conn. Agr. Exp. Sta. Bull. 188.
, and R. J. GARBER, 1919. Synthetic production of high-protein
corn in relation to breeding. Jour. Am. Soc. Agron., 11 : 309-318.
, and R. J. GARBER, 1927. Breeding Crop Plants, Ed. 2. McGraw-
Hill Book Company, Inc., New York.
, FRED GRIPFEE, F. J. STEVENSON, and A. P. LTINDEN, 1928. Corre-
lated studies in oats of the inheritance of reaction to stem rust and smuts
and of other differential characters. Jour. Agr. Research, 36 : 437-457.
, and H. V. HARLAN, 1920, The inheritance of the length of inter-
node in the rachis of the barley spike. U.S. Dept. Agr. Bull. 869.
, and I. J. JOHNSON, 1939. The breeding of improved selfed lines of
corn. Jour. Am. Soc. Agron., 31: 710-724.
f M. B. MOORE, arid E. C. STAKMAN, 1939. Studies of inheritance
in crosses between Bond, A vena byzantina, and varieties of A. saliva.
Minn. Agr. Exp. Sta. Tech. Bull. 137.
, J. H. PARKER, and CARL KTJRTXWEIL, 1920. Genetics of rust
resistance in crosses of Tnticum vulgare with varieties of T. durum and
T. dicoccum. Jour. Agr. Research, 19: 523-542.
, E. C. STAKMAN, and O. S. AAMODT, 1925. Inheritance in wheat of
resistance to black stem rust. Phytopathology, 15 : 371-387.
, E. C. STAKMAN, FRED GRIFFEE, and J. J. CHRISTENSEN, 1923.
Reaction of barley varieties to Helminthosporium sativum. Minn* Agr.
Exp. Sta. Tech. Bull. 21.
, E. C. STAKMAN, FRED GRIFFEE, and J. J. CHRISTENSEN, 1924.
Reaction of selfed lines of maize to Ustilago zeae. Phytopathology, 14 :
268-280.
HENKEMEYEF^ A., 1915. Untersuchungen liber die Spaltungen von Weizen-
bastarden in der F% and F$ Generationen. Jour, fur experimentalle
Landwirtschaft 63 : 97-124.
HENRY, A. W., 1930. Inheritance of immunity from flax rust. Phyto-
pathology, 20:707-721.
HERIBERT-NILSSON, N., 1916. Populationsanalysen und Erblichkeits-
vereuche fiber die Selbssterilitat, Selbstfertilitat, und Sterilitat bei dem
Roggen. Zeit. fur Pflanzenzucht 4 : 1-44.
LITERATURE CITATIONS 389
, 1919. Ragforadlingens Methodik och Principer. Saertryk af
Nordisk Jorbrugsforskning.
1921. Selektive Verschlebung der Gametenfrequenz in einer Kreu-
zungspopulation von Roggen. Hereditas, 2: 364-369.
HEYNE, E. G., and A. M. BRUNSON, 1940. Genetic studies of heat and
drought tolerance in maize. Jour. Am. Soc. Agron., 32: 803-814.
, and H. H. LAUDE, 1940. Resistance of corn seedlings to high tem-
peratures in laboratory tests. Jour. Am. Soc. Agron., 32: 116-126.
HITCHCOCK, A. S., 1935. Manual of the grasses of the United States. U.S.
Dept. Agr. Misc. Pub. 200.
HOLBERT, J. R., 1924. Anchorage arid extent of corn root system. Jour.
Agr. Research, 27: 71-78.
, and W. L. BTJRLISON, 1928. Corn breeding for resistance to cold
yields good results. U.S. Dept. Agr. Yearbook, pp. 227-229.
HOOVER, M. M., 1932. Inheritance studies of the reaction of selfcd lines of
maize to smut (Ustilago zeae). West Va. Agr. Exp. Sta. Bull. 253.
HOPKINS, C. G., 1899. Improvement in the chemical composition of the
corn kernel. III. Agr. Exp. Sta. Bull, 55.
HOR, K. S., 1924. Interrelations of genetic factors in barley. Genetics, 9:
151-180.
HOWARD, A., and G. L. C. HOWARD, 1915. On the inheritance of
some characters in wheat. II. Mem. Dept. Agr. India Bot. Ser. 7:
273-285.
HUMPHREY, L. M., 1940. Effects of inbreeding cotton with special reference
to staple length and lint percentage. Ark. Agr. Exp. Sta. Bull. 387.
, and A. V. TULLBR, 1938. Improvement in the technique of cotton
hybridization. Ark. Agr. Exp. Sta. Bull. 359.
HUNTER, H., and H. M. LEAKE, 1933. Recent Advances in Agricultural
Plant Brooding. The Blakiston Company, Philadelphia.
HUNTER, J. W., H. H. LAUDE, and A. M. BRUNSON, 1936. A method for
studying resistance to drought injury in inbred lines of maize. Jour.
Am. Soc. Agron., 28: 694-698.
HUTCHINS, A. E., 1938. Heterosis in the cucumber. Proc. Amer.Soc.HorL
Sri., 36:660-664.
IMMER, F. R., 1927. The inheritance of reaction to Ustilago zeae in rnaize.
Minn. Agr. Exp. Sta., Tech. Bull. 51.
~, 1930. Formulae and tables for calculating linkage intensities.
Genetics, 16: 81-98.
-, 1934. Varietal competition as a factor in yield trials with sugar
beets. Jour. Am. Soc. Agron., 26: 259-261.
, 1941. Relation between yielding ability and homozygosis in
barley crosses. Jour. Am. Soc. Agron., 33: 200-206.
, H. K. HAYES, and LfiRoy POWERS, 1934. Statistical determination
of barley varietal adaptation. Jour. Am. Soc. Agron., 26: 403-4J9.
, and S. M. RALEIGH, 1933. Further studies on size and shape of
plot in relation to field experiments with sugar beets. Jour. Agr.
Research, 47 : 591-598.
390 METHODS OF PLANT BREEDING
IVANOFF, S. S., and A. J. HIKER, 1936. Resistance to bacterial wilt of
inbred strains and crosses of sweet corn, Jour. Agr. Research, 53 : 927-
954.
JENKIN, T. J., 1924. The artificial hybridization of grasses. Welsh Plant
Breeding Station Bull., Series H, No. 2.
, 1931a. Self-fertility in perennial rye-grass (LoUum perenne L.).
Welsh Plant Breeding Station Bull, Series H, No. 12, pp. 100-119.
, 19316. Fertility in plants of the genus Phleum. Welsh Plant Breed-
ing Station Bull., Series II, No. 12, pp. 148-159.
1931e. The method of selection, breeding and strain building in
grasses. Imp. Bur. of Plant Genetics: Herbage Plants Bull. 3.
JENKINS, M. T., 1929. Correlation studies with inbred and cross-bred
strains of maize. Jour. Agr. Research, 39: 677-721.
, 1932. Differential resistance of inbred and crossbred strains of
corn to drought and heat injury. Jour. Am. Soc. Agron., 24: 504-506.
, 1934. Methods of estimating the performance of double crosses in
corn. Jour. Am. Soc. Agron., 26: 199-204.
, 1935. The effect of inbreeding and selection within inbred lines of
maize upon the hybrids made after successive generations of selling.
Iowa State College \Iour. Sci., 9: 429-450.
~ , 1936. Corn improvement. U.S. Dept. Agr. Yearbook, pp. 455-522.
- , 1940. The segregation of genes affecting yield of gram in maize.
Jour. Am. Soc. Agron,, 32: 55-63.
arid A. M. BRUNRON, 1932. Methods of testing inbred lines of
make in crossbred combinations. Jour. Am. Soc. Agro?i., 24: 523-530.
JODON, N. E., 1938. Experiments on artificial hybridization of rice. Jour.
Am. Soc. Agron., 30: 294-305.
JOHANNWEN, W., 1903. Ucber Krblichkeit in Populationen und in reinen
Linien. Gustav Fischer, Jena.
, 1909. Elemente der exakten Erblichkeitslehre. Gustav Fischer,
Jena.
JOHNSON, I. J., 1932. Correlation studies with strains of flax with particular
reference to the quantity and quality of the oil. Jour. A m. Soc. Agron.,
24: 537-544.
, and H. K. HAYES, 1936. The combining ability of inbred lines of
Golden Bantam sweet corn. Jour. Am, Soc. Agron., 28: 246-252.
, and H. K. HAYES, 1938. The inheritance of pericarp tenderness in
sweet corn. Jour Am. Soc. Agron., 30: 220-231.
and H. K. HAYKR, 1940. The value in hybrid combinations of
inbred lines of corn selected from single crosses by the pedigree method
of breeding. Jour. Am. Soc. Agron., 32: 479-485.
JOHNSON, T., and M ABO ABET NEWTON, 1940. Mendelian inheritance of
certain pathogenic characters of Puccinia graminis tritici. Canadian
Jour. Research, C. 18: 599-611.
JONES, D. F., 1916. Natural cross-pollination in the tomato. Science,
43 : 509-510.
, 1917. Dominance of linked factors as a means of accounting for
heterosis. Genetics, 2 : 466-479.
LITERATURE CITATIONS 391
, 1918. The effects of inbreeding and crossbreeding upon develop-
ment. Conn. Agr. Exp. Sta. Bull. 207.
, 1920. Selection in self fertilized lines as the basis for corn improve-
ment. Jour. Am. Soc. Agron., 12: 77-100.
, 1924. The origin of flint and dent corn. Jour. Heredity, 16: 417-
419.
, 1939. Continued inbreeding in maize. Genetics, 24: 462-473.
-, and W. R. SINGLETON, 1935. The improvement of naturally
cross-pollinated plants by selection in self-fertilized lines. II. The
testing and utilization of inbred strains of corn. Conn, Agr. Exp. Sta.
Bull. 375.
JONES, H. A., 1932. Vegetable breeding at the University of California.
Proc. Amer. Soc. Hort. Science, 29: 572-581.
, and S. L. EMSWELLER, 1933. Methods of breeding onions. II il-
gardia, 1 : 625-042.
JONES, J. W., 1936. Improvement in rice. U.S. Dept. Agr. Yearbook,
pp. 415-454.
JONES, L. R., and J. C. OILMAN, 1915. The control of cabbage yellows
through disease resistance. Wis. Agr. Exv. Sta. Res. Bull. 38.
JORGENSON, Locus, and H. E. BREXVKAKER, 1927. A comparison of selfed
lines of corn and first generation crosses between them. Jour. Am.
Soc. Agron., 19: 819-830.
KEARNEY, T. 11., 1923. Self fertilization and cross fertilization in Pima
cotton. U.S. Dept, Agr. Bull. 1134.
KEEBLE, FREDERICK, and C. PELLEW, 1910. The mode of inheritance of
stature arid of time of flowering in peas (Pisum sativum). Jour. Genetics,
1 : 47-56.
KEMPTON, J. K., 1924. Inheritance of proterogyny in maize. Am.
Naturalist, 58: 182-187.
KEZEK, ALVIN, and BREEZE BOYACK, 1918. Mcndelian inheritance in
wheat and barley crosses. Colo. Agr. Exp. Sta. Bull. 249.
KIESWELBAOH, T. A., 1918. Studies concerning the elimination of experi-
mental error in comparative crop tests. Ncbr. Agr. Exp. Sta. Res.
Bull. 13.
, 1922. Corn investigations. Ncbr. Agr. Exp. Sta. Res. Bull. 20.
, 1923. Competition as a source of error in comparative corn yields.
Jour-. Am. Soc. Agron., 16: 199-215.
, 1930. The use of advanced-generation hybrids as parents of double
cross seed corn. Jour. Am. Soc. Agron., 22: 614-626.
-, and 11, M. WEIHING, 1933. Effect of stand irregularities upon the
acre yield and plant variability in corn. Jour. Agr. Research, 47:
399-416.
KIHARA, H., and I. NISIIIYAMA, 1932. The genetics and cytology of certain
cereals, III. Different compatibility in reciprocal crosses of A vena with
special reference to tetraploid hybrids between hexaploid and diploid
species. Japan. Jour. Bot., 6: 245-305.
KIRK, L. E., 1927. Self-fertilization in relation to forage crop improvement.
Sci. Agr., 8: 1-40.
392 METHODS OF PLANT BREEDING
, 1930. Abnormal seed development in sweet clover species crosses.
A new technique for emasculating sweet clover flowers. Sci. Agr., 10 :
321-327.
, 1 932. Methods employed in the breeding of biennial sweet clover,
Melilotus, and brief notes on the breeding of Lucerne, Medicago saliva,
bronie grass, Bromus inermis, and slender wheat grass, Agropyron
tenerum. Imp. Bur. Plant Genetics: Herbage Plants. Bull. 7.
1933. The progeny test and methods of breeding appropriate to
certain species of crop plants. Am. Naturalist, 67; 515531.
KOSTOFF, DONOHO, 1937. Cytological studies on certain progenies of the
hybrid Triticum timopheevi X Triticum persicum. Cytologia, Fujii
Jubilee Vol., pp. 262-277.
LARSON, A. H., R. B. HARVEY, and JOHN LARSON, 1936. Length of the
dormancy period in cereal seeds. Jour. Agr. Research, 62: 811-836.
LEONARD, W. H., 1940. Inheritance of fertility in the lateral spikelets of
barley. Ph.D. thesis. University of Minnesota.
LEVY, E. B., 1933. Strain testing and strain building. Technique employed
in grassland research in New Zealand. Imp. Bur. of Plant Genetics:
Herbage Plants. Bull. 11, pp. 6-16.
LILIENFELD, F., and H. KIHARA, 1934. Genomarialyse bei Triticum und
Aegilops. V. Triticum timopheevi Zhuk. Cytologia, 6: 87-122.
LINDSTROM, E. W., 1931. Genetic tests for linkage between row number
genes and certain qualitative genes in maize. Iowa Agr. Exp. Sta.
Res. Bull. 142.
, 1935. Genetic experiments on hybrid vigor in maize. Am.
Naturalist, 69: 311-322.
LOVE, H. H., and W. T. CRAIG, 1918a. Methods used and results obtained
in cereal investigations at the Cornell station. Jour. Am. 8oc. Agr on.,
10:145-157.
, and W. T. CRAIG, 19186. The relation between color and other
characters in certain Avena crosses. A m. Naturalist, 52 : 369-383.
and W. T. CRAIG, 1924. Methods now in use in cereal breeding
and testing at the Cornell Agricultural Experiment Station. Jour. Am.
Soc. Agron., 16: 109-127.
, and G. P. McRosTiB, 1919. The inheritance of hull-lessness in oat
hybrids. Am. Naturalist, 63: 5-32.
LOVE, R. M., 1938. Somatic variation of chromosome numbers in hybrid
wheats. Genetics, 23 : 517-523.
LUCKWILL, L. C., 1937. Studies on the inheritance of physiological char-
acters. IV. Hybrid vigour in the tomato. Part 2. Manifestation of
hybrid vigour during the flowering period. Ann. Bot. (N.S.), 1: 379-
408.
MACATJLEY, T. B., 1928. The improvement of corn by selection and plot
inbreeding. Jour. Heredity, 19: 57-72.
MACINDOE, S. L., 1931. Stem rust of wheat. Agr. Gazette N.S. W., 42 : 475-
484.
MAINS, B. B., 1931. Inheritance of resistance to rust, Puccinia sorghi, in
maize. Jour. Agr. Research, 43: 419-430.
LITERATURE CITATIONS 393
MANGBLSDORF, P. C., 1926. The genetics and morphology of some endo-
sperm characters in maize. Conn. Agr. Exp. Sta. Bull. 279.
, and G. S. FRAPS, 1931. A direct quantitative relationship between
vitamin A in corn and the number of genes for yellow endosperm.
Science, 73:241-242.
-, and R. G. REEVES, 1939. The origin of Indian corn and its rela-
tives. Texas Agr. Exp. Sta. Bull. 574 (monograph).
MARSTON, A. R., 1930. Breeding corn for resistance to the European corn
borer. Jour. Am. Sac. Agron., 22: 986-992.
MCFADDEN, E. S., 1 930. A successful transfer of emmer characters to vulgare
wheat. Jour. Am. Soc. Agron., 22: 1020-1034.
MERCER, W. B., and A. D. HALL, 1911. The experimental error of field
trials. Jour. Agr. Sri., 4: 107-127.
MEYERS, M. T., L. L. HUBER, C. R. NEISWANDER, F. D. RICKEY, and G. H.
STRING FIELD, 1937. Experiments on breeding corn resistant to the
European corn borer. U.S. Dept. Agr, Tech. Bull. 583.
MILLER, H. ,!., 1936. Out of the Night. Victor Gollancz, Ltd., London.
MONTGOMERY, E. G., 1909. Experiments with corn. Nebr. Agr. Exp. Sta.
Bull. 112.
MORSE, W. J., and J. L. CARTTER, 1937. Improvement in soybeans. U.S.
Dept. Agr. Yearbook, pp. 1154-1189.
MURPHY, H. C., T. R. STANTON, and F. A. COFFMAN, 1936. Hybrid selec-
tions of oats resistant to smuts and rusts. Jour. Am. Soc. Agron., 28:
370-373.
MURPHY, R. P., 1941. Convergent improvement with four inbred lines of
corn. Part B of Ph.D. thesis. University of Minnesota,
MYERS, W. M., 1936. A correlated study of the inheritance of seed size and
botanical characters in the flax cross, Redwing X Ottawa 770B. Jour.
Am. Soc. Agron., 28: 623-635.
, 1937. The nature and interaction of genes conditioning reaction to
rust in flax. Jour. Agr. Research, 66: 631-666.
and LER.OY POWERS, 1938. Meiotic instability as an inherited
character in varieties of Triticum aestivum. Jour. Agr. Research, 66:
441-452.
NEAL, N. P., 1935. The decrease in yielding capacity in advanced genera-
tions of hybrid corn. Jour. Am. Soc. Agron., 27: 666-670.
NEBEL, B. R., and M. L. RUTTLE, 1938. Colchicine and its place in fruit
breeding. New York (Geneva) Agr. Exp. Sta. Circ. 183.
NEWMAN, L. H., 1912. Plant breeding in Scandinavia. Canadian Seed
Growers' Association, Ottawa, Canada.
NILSSON-EHLE, H., 1909. Kreuzungsuntersuchungen an Hafer und Weizen.
Lunds Univ. Arsskr. N.F. Afd. 2. Bd. 5. Nr. 2, pp. 1-122.
, 191 la. Kreuzungsuntersuchungen an Hafer und Weizen. Lunds
Univ. Arsskr. N.F. Afd. 2, Bd. 7. Nr. 6, pp. 3-84.
-, 19116. Ueber Falle spontanen Wegf aliens eines Hemmungsfaktors
beim Hafer. Zeitschr. fur Induk. Abstamm. u. Vererb., 6: 1-37.
NILSSON-LEISSNEB, GUNNAR, 1925. Beitrage zur Genetik von Triticum
spelta und Triticum vulgare. Hereditas, 7 : 174.
394 METHODS OF PLANT BREEDING
, 1927. Relation of selfed strains of corn to F\ crosses between them.
Jour. Am, Soc. Agron., 19: 440-454.
NISHIYAMA, I., 1929. The genetics and cytology of certain cereals. I.
Morphological and cytological studies on triploid, pentaploid, and
hcxaploid Avena hybrids. Japan. Jour, Genetics, 6: 1-48.
NOLL, C. P., 1927. Mechanical operations of small grain breeding. Jour.
Am. Soc. Agron., 19: 713-721.
NOWOSAD, F. S., and R. M. MAOVICAR, 1940. Adaptation of the "picric
acid test" method for selecting HCN free lines in sudan grass. Sci.
Agr., 20: 566-569.
ORTON, W. A., 1913. The development of disease resistant varieties of
plants. IV. Conference Internationale de G6n6tique, Paris, 1911.
PAINTER, R. H., E. T. JONES, C. O. JOHNSON, and J. H. PARKER, 1940.
Transference of Hessian fly resistance and other characteristics of
Marquillo spring wheat to winter wheat. Kan. Agr. Exp. 8ta. Tech.
Bull, 49.
PAN, C. L., 1940. A genetic study of mature plant resistance in spring wheat
to black stern rust, Puccinia graminis tritici, and reaction to black chaff,
Bacterium translucens var. undulosum. Jour. Am. Soc. Agron., 32:
107-115.
PARKER, W. H., 1914. Lax and dense eared wheats. Jour, Agr. Sci., 6:
371-386.
PEARSON, KARL, 1924. Tables for Statisticians and Biometricians, Part I,
Ed. 2. Cambridge University Press, London.
PEARSON, O. H., 1932. Breeding plants of the cabbage group. Calif. Agr.
Exp. Sta. Bull. 532.
PELSHENKE, P., 1930. Beitriige zur Bestimmung der Backfahigkeit von
Weizen und Weizemnehlen. Arch. Pfianzcnbau, 5: 108-151.
, 1933. A short method for the determination of gluten quality in
wheat. Cereal Chemistry, 10 : 90-96.
PERCIVAL, J., 1921 . The wheat plant: A monograph. Gerald Duckworth &
Company, Ltd., London.
PETERSON, R. F., 1934. Improvement of rye through inbreeding. Sci.
Agr., 14: 651-668.
, T. JOHNSON, and MARGARET NEWTON, 1940. Varieties of Triticum
vulgare practically immune in all stages of growth to stem rust. Science,
91:313.
PHILP, JAMES, 1933. The genetics and cytology of some interspecific
hybrids of Avena. Jour. Genetics, 27: 133-179,
PIETERH, A. J., and E. A. HOLLOWELL, 1937. Clover improvement. U.S.
Dept. Agr. Yearbook, pp. 1190-1214.
PORTER, D. R., 1933. Watermelon breeding. Hilgardia, 1: 585-624.
POWERS, LsRoY, 1932, Cytologic and genetic studies of variability of
strains of wheat derived from interspecific crosses. Jour. Agr. Research.
44:797-831.
, 1934. The nature and interaction of genes differentiating habit c
growth in a cross between varieties of Triticum vulgare. Jour. Agr.
Research, 49 : 573-605.
LITERATURE CITATIONS 395
-, 1936. The nature of the interaction of genes affecting four quan-
titative characters in a cross between Hordeum deficiens and vulgare.
Genetics, 21 : 398-420.
-, and LEE HINES, 1933. Inheritance of reaction to 'stem rust and
barbing of awns in barley crosses. Jour. Agr. Research, 46: 1121-1129.
PRELL, II., 1921. Das Problem der Unbefruchtbarkeit. Naturwiss t 20:
440-446.
QUISENBERRY, K. S., and J. A. CLARK, 1933. Inheritance of awn develop-
ment in Sonera wheat crosses. Jour. Am. Soc. Agron., 25 : 482-492.
RASMUSSON, J., 1935. Studies on the inheritance of quantitative characters
in Pisum. Hereditas, 20: 161-180.
REED, G. M., 1935. Physiological specialization of the parasitic fungi.
Botanical Review, 1: 119-137.
, 1940. Physiologic races of oat smuts. Am. Jour. BoL, 27:
135-143.
REID, D. A., 1938. A study of the inheritance of seedling and mature plant
reaction to Puccinia graminis tritici in a cross of Wisconsin 38 X
Peatland barley. M.S. thesis, University of Minnesota.
RICKEY, F. D., 1922. The experimental basis for the present status of corn
breeding. Jour. Am. Soc. Agron., 14: 117.
, 1924. Effects of selection on the yield of a cross between varieties
of corn. U.S. DepL Agr. Bull. 1209.
, 1927. The convergent improvement of selfed lines of corn. Am.
Naturalist, 61 : 430-449.
, and L. S. MAYER, 1925. The productiveness of successive genera-
tions of self-fertilized lines of corn and of crosses between them. U.S.
Dept. Agr. Bull. 1354.
, and G. F. SPRAGUE, 1931. Experiments on hybrid vigor and con-
vergent improvement in corn. U.S. Dept. Agr. Tech. Bull. 267.
-, G. H. STRINGFIELD, and G. F. SPRAGUE, 1934. The loss in yield
that may be expected from planting second generation double-crossed
seed corn. Jour. Am. Soc. Agron., 26: 196-199.
RIDER, PAUL R., 1939. An Introduction to Modern Statistical Methods.
John Wiley & Sons, Inc., New York.
RIEMAN, G. H., 1939. The importance of disease resistant varieties in a
programme of vegetable seed production. Annual Rept. Canadian Seed
Growers' Assoc., Ottawa.
RILEY, H. P., 1934. A further test showing the dominance of self-fertility to
self-sterility in shepherd's purse. Am. Naturalist, 68: 60-64.
, 1936. The genetics and physiology of self -sterility in the genus
Capsella. Genetics, 21 : 24-39.
ROBERTS, W. L., aDd K. P. LINK, 1937. Determination of coumarin and
mclilotic acid. Ind. and Eng, Chem., 9: 438-441.
ROBERTSON, D. W., 1933. Inheritance in barley. Genetics, 18: 148-158.
, G. A. WIEBE, and F. R. IMMER, 1941. A summary of linkage studies
in barley. Jour. Am. Soc. Agron., 33: 47-64.
ROSA, J. T., 1927. Results of inbreeding melons, Proc. Amer, Soc, Hort,
ScL } 24:79-84,
396 METHODS OF PLANT BREEDING
SABOE, L. C., and H. K. HAYES, 1941. Genetic studies of reactions to smut
and firing in maize by means of chromosomal interchanges. Jour. Am.
Soc. Agron., 33 : 463-470.
SALMON, S. C.,' 1931. An instrument for determining the breaking strength
of straw and a preliminary report on the relation between breaking
strength and lodging. Jour. Agr. Research, 43: 73-82.
, 1 938. Generalized standard errors for evaluating bunt experiments
with wheat. Jour. Am. Soc. Agron., 30: 647-663.
SANSOME, F. W., and J. PHILP, 1939. Recent Advances in Plant Genetics,
Ed. 2. The Blakiston Company, Philadelphia.
SAUNDEBS, H. A., and S. HUMPHRIES, 1928. The Saunders test. Jour.
Inst. Agr. Bot., 2 : 34.
SCOTT, G. W., 1932. Inbreeding studies with Cucumis melo. Proc. Amer.
Soc. Hort. Sci., 29: 485.
SHAMEL, A. D., and C. S. POMEROY, 1932. Bud variations in apples. Jour.
Heredity, 23: 173-180, 213-220.
, L. B. SCOTT, and C. S. POMKROY, 1918a. Citrus-fruit improvement:
A study of bud variation in the Washington navel orange. U.S. Dept.
Agr. Bull. 623.
, L. B. SCOTT, and C. S. POMEROY, 19186. Citrus-fruit improvement:
A study of bud variation in the Valencia orange. U.S. Dept. Agr. Bull.
624.
-, L. B. SCOTT, and C. S. POMKROY, 1918c. Citrus-fruit improvement:
A study of bud variation in the Marsh grapefruit. U.S. Dept. Agr.
Bull. 697.
SHAW, F. J. F., A. R. KHAN, and M. A LAM, 1931. Studies in Indian oil
seeds. V. The inheritance of characters in Indian linseed. Indian
Jour. Agr. Sci., 1 : 1-57.
SHULL, G. H., 1909. A pure line method of corn breeding. Am. Breeders
Assoc. Rept. 5: 51-59.
, 1910. Hybridization methods in corn breeding. Am. Breeders
Magazine 1 : 98-107.
1914. Duplicate genes for capsule form in Bursa bursa-pastoris.
Zietschr. fur Induk. Abstamm. u. Vererb., 12: 97149.
SMITH, D. C., 1934. Correlated inheritance in oats of reaction to diseases
and other characters. Minn. Agr. Exp. Sta. Tech. Bull. 102.
SMITH, II. H., 1939. The induction of polyploidy in Nicotiarui species and
species hybrids. Jour. Heredity, 30: 291-306.
SMITH, L. H., and A. M. BRUNSON, 1925. An experiment in selecting corn
for yield by the method of the ear-row breeding plot. 111. Agr. Exp.
Sta. Bull. 271.
SNEDECOR, G. W., 1940. Statistical Methods, Ed. 3. Iowa State College
Press, Ames, Iowa.
SNELLING, R. O., R. A. BLANCH ARD, and J. H. BIGGER, 1940. Resistance
of corn strains to the leaf aphid, Aphis maidis Fitch. Jour. Am. Soc.
Agron., 32: 371-381.
, and R. G. DAHMS, 1937. Resistant varieties of sorghum and corn in
relation to chinch bug control in Oklahoma. Okla. Agr. Exp. Sta. Bull,
232,
LITERATURE CITATIONS 397
SPILLMAN, W. J., 1909. The hybrid wheats. Wash. Agr. Exp. Sta. Bull. 89.
SPRAGUE, G. F., 1936. Hybrid vigor and growth rate in a maize cross and
its reciprocal. Jour. Agr. Research, 63: 81^-830.
, and A. A. BRYAN, 1941. The segregation for genes affecting yield
prepotency, lodging and disease resistance in F$ and ^4 lines of corn.
Jour. Am. Soc. Agron., 33: 207-214.
STADLEB, L. J., 1928. Mutations in barley induced by x-rays and radium.
Science, 68: 186-187.
, 1929. Chromosome numbers and the mutation rate in Avena and
Triticum. Proc. Nat. Acad. Sci., 16: 876-381.
STAKMAN, E. C., 1914. A study of cereal rusts: physiological races. Minn.
Agr. Exp. Sta. Bvll. 138.
, L. O. KUNKEL, A. J. RlKER, J. H. CRAIGIE, H. A. RODENHISER,
and J. J. CHRISTENSEN, 1940. The genetics of pathogenic organisms,
pp. 9-17, 22-27, 46-56, and 66-90. Publication of American Associa-
tion for the Advancement of Science, No. 12. Science Press.
-, M. N. LEVINE, J. J. CHRISTENSEN, and K. ISENBECK, 1935. Die
Bestimmung physiologischer Rassen pflanzenpathogener Pilze. Nova
Acta Leopoldina (N.S.), 3: 281-336.
STANFORD, E. H., and F. N. BRIGGS, 1940, Two additional factors for
resistance to mildew in barley. Jour. Agr. Research, 61: 231-236.
STANTON, T. R., 1933. Navarro oats. Jour. Am. Soc. Agron., 26 : 108-1 12.
, 1936. Superior germ plasm in oats. U. S. Dept. Agr. Yearbook,
pp. 347-414.
, and H. C. MURPHY, 1933. Oat varieties highly resistant to crown
rust and their probable value. Jour. Am. Soc. Agron., 26 : 674683.
, H. C. MURPHY, F. A. COFFMAN, and H. B. HUMPHREY, 1934.
Development of oats resistant to smuts and rusts. Phytopathology, 24 :
165-167.
, G. M. REED, and F. A. COFFMAN, 1934. Inheritance of resistance
to loose smut and covered smut in some oat hybrids. Jour. Agr.
Research, 48: 1073-1083.
STAPLEDON, R. G., 1931. Self- and cross-fertility and vigour in cocksfoot
grass (Dactylis glomerata L.). Welsh Plant Breeding Station Bull.,
Series H, No. 12, pp. 161-180.
STEPHENS, J. C., and J. R. QUINSY, 1933. Bulk emasculation of sorghum
flowers. Jour. Am. Soc. Agron., 26: 233-234.
STEVENSON, F. J., 1928. Natural crossing in barley. Jour. Am. Soc.
Agron., 28: 1193-1196.
STEVENSON, T. M., and J. S. CLAYTON, 1936. Investigations relative to the
breeding of coumarin-free sweet clover, Melilotus. Can. Jour. Research,
14: 153-165.
, and W. J. WHITE, 1940. Investigations concerning the coumarin
content of sweet clover. Sci. Agr., 21 : 18-28.
STEWART, GEORGE, 1926. Correlated inheritance in wheat. Jour. Agr.
Research, 33: 1163-1192.
STURTEVANT, E. L., 1899. Varieties of corn, U.S. Dept. Agr. Office Exp.
Sta. Butt. 57.
398 METHODS OF PLANT BREEDING
SUMMERBY, R., 1934. The value of preliminary uniformity trials in increas-
ing the precision of field experiments. MacDonald College (McGill
University) Tech. Bull. 15.
SUNESON, C. A., 1937. Emasculation of wheat by chilling. Jour. Am. Soc.
Agron., 29: 247-249.
SURFACE, F. M., 1916. Studies on oat breeding. III. On the inheritance
of certain glume characters in the cross of Avenafatua X A. saliva var.
Kherson. Genetics, 1 : 252-286.
TAMMES, TINE, 1928. The genetics of the genus Linum. Bibliog. Genetica
4; 1-36.
TEDIN, OLOF, 1931. The influence of systematic arrangement upon the
estimate of error in field experiments. Jour. Agr. Sci., 21: 191-208.
TIPPETT, L. H. C., 1927. Tracts for Computers. No. XV. Random
Sampling Numbers. Cambridge University Press, London.
TISDALE, W. H., 1916. Relation of soil temperature to infection of flax by
Fusarium lini. Phytopathology, 6: 412-413.
, 1917. Flax wilt. A study of the nature and inheritance of wilt
resistance. Jour. Agr. Research, 11: 573-605.
TORRIE, J. H,, 1939. Correlated inheritance in oats of reaction to smuts,
crown rust, stem rust, and other characters. Jour. Agr. Research, 69:
783-804.
TYSDAL, H. M., 1941. Effect of self-pollination upon seed and forage yield
in alfalfa. Unpublished.
, and J. R. GAEL, 1938. A new method for alfalfa emasculation.
Jour. Am. Soc. Agron., 32: 405-407.
and T. A. KIEBSELBACH, 1939. Alfalfa nursery technic. Jour.
Am. Soc. Agron., 31: 83-98.
U.S. Dept. Agr. Yearbook of Agriculture, 1936.
U.S. Dept. Agr. Yearbook of Agriculture, 1937.
VALLE, O., 1931. Untersuchungen liber die Selbststerilitat und Selbstfer-
tilitat des Timothes (Phleurn pratense L.) und die Einwirkund der
Selbstbefruchtuiig auf die Nachkommenschaft. Eine ziichtungs-
methodische Studie. Acta Agralia Fennica, 24: 1-257.
VAVILOV, N. L, 1939. Genetics in the U.S.S.R. Chron. BoL, 6: 14-15.
, and E. S. KOUZNETSOV, 1921. On the genetic nature of winter and
spring varieties of plagts. Izvestii Agron. Facul'tete Saratovskogo Univ.
(Bull. Agron. Faculty Saratov Univ.), pp. 1-25. (In Russian; English
summary, pp. 23-25.)
VILMORIN, Louis DE, 1856. Note sur la creation d'une nouvelle race de
betterave a sucre, pp. 25-29. Notices sur 1 'amelioration des plantes,
by Louis Levque de Vilmorin and Andre* Lev^que de Vilmorin. Vil-
morin- Andrieux, Paris.
VOGEL, O. A., 1938. The relation of lignification of the outer glume to
resistance to shattering in wheat. Jour. Am. Soc. Agron. , 30 : 599-603.
WADE, B. L., 1937. Breeding and improvement of peas and beans. U.S.
Dept. Agr. Yearbook, pp. 251-282.
WALKEK, J. C., 1935. Types of disease resistance. Zesde Internationaal
Botanisch Congress Proc. 2 : 200-208,
LITERATURE CITATIONS 399
WALLACE, H. A,, and E. N. BRESSMAN, 1928. Corn and Corn Growing.
John Wiley & Sons, Inc., New York.
, and G. W. SNEDECOR, 1931, Correlation and machine calculation.
Revised. Iowa State College Pub. 30.
WANG, K. W., 1939. Some phases of heterosis in corn. Ph.D. thesis.
University of Minnesota.
WARMKE, H. E., and A. F. BLAKESLEE, 1939. Induction of simple and
multiple polyploidy in Nicotiana by colchicine treatment. Jour.
Heredity, 30: 419-432.
W ATKINS, A. E., 1940. The inheritance of glume shape in Triticum. Jour.
Genetics 39 : 249-264.
, and SYDNEY ELLERTON, 1940. Variation and genetics of the awn
in Triticum. Jour. Genetics, 40 : 243-270.
WEIBEL, R. O., and K. S. QUISENBERRY, 1941. Field versus controlled freez-
ing as a measure of cold resistance of winter wheat varieties. Jour.
Am. Soc. Agr on., 33: 336-343.
WELLHATJSEN, E. J., 1937. Genetics of resistance to bacterial wilt in maize.
Iowa Agr. Exp. Sta. Res. Bull. 224.
WELSH, J. N., 1931. The inheritance of stem rust reaction and lemma
colour in oats. Sci. Agr. 12 : 209-242.
, 1937. The synthetic production of oat varieties resistant to race 6
and certain oilier physiologic races of oat stem rust. Canadian Jour.
Research, 16 : 58-69.
WEXELSEN, H., 1934. Quantitative inheritance and linkage in barley.
Hereditas, 18 : 307-348.
WHALEY, W. G., 1939a. A development analysis of heterosis in Lycoper-
sicum. I. The relation of growth rate to heterosis. Am. Jour. Botany,
26:609-616.
, 19396. A development analysis of heterosis in Ly coper si cum. II.
The role of the apical meristem in heterosis. Am. Jour. Botany, 26:
682-690.
WHITAKER, T. W., and I. C. JAGGER, 1937. Breeding and improvement of
cucurbits. U.S. Dept. Agr. Yearbook, pp. 207-232.
WIEBE, G. A., 1935. Variation and correlation in grain yields among 1500
wheat nursery plots. Jour. Agr. Research, 50: 331-357.
WIENER, W. T. G. ? 1937. The Canadian Seed Growers' Association. Her-
bage Reviews, 5: 14-18. Imp. Bur. Plant Genetics: Herbage Plants.
WILLIAMS, C. G., 1905. Pedigreed seed corn. Ohio Agr, Exp. Sta.
Circ. 42.
, 1907. Corn breeding and registration. Ohio Agr. Exp. Sta. Circ.
66.
, and F. A. WELTON, 1915. Corn experiments. Ohio Agr. Exp. Sta.
Bull. 282.
WILLIAMS, R. D,, 1931 a. Self- and cross-sterility in red clover. Welsh
Plant Breeding Station Bull, Series H, No. 12, pp. 181-208.
, 19316. Self- and cross-sterility in white clover. Welsh Plant
Breeding Station Bull., Series H. No. 12, pp. 209-216.
400 METHODS OF PLANT BREEDING
, 1931c. Methods and techniques of breeding red clover, white
clover and lucerne. Imp. Bur. Plant Genetics: Herbage Plants Bull. 3,
pp. 46-76.
WRIGHT, SEWALL, 1921. 'Systems of mating. V, General considerations.
Genetics, 6: 167-178.
Wu, SHAO-KWEI, 1939. The relationship between the origin of selfed lines
of corn and their value in hybrid combination. Jour. Am. Soc. Agron.,
31 : 131-140.
YASTTDA, S., 1934. Physiological research on self-incompatibility in Petunia
violacea. Bull. Imp. Col. Agri. and Forestry Morioka, 20: 1095.
YATES, F., 1933. The analysis of replicated experiments when the field
results are incomplete. Empire Jour. Exp. Agr., 2: 129-142.
, 1936. A new method of arranging variety trials involving a large
number of varieties. Jour. Agr. Sci.j 26: 424-455.
, 1939. The recovery of inter-block information in variety trials
arranged in three-dimensional lattices. Ann. Eugenics, 9: 136-156.
GLOSSARY
Allele, allel, allelomorph: adjective forms: allelic, allelomorph ic. One of a
pair, or of a series of factors, that occur at similar loci of homologous
chromosomes and for this reason are inherited in alternative pairs.
One alternative form of a gene.
Aleurone. The protein grains found in the endosperm of ripe seeds.
Aleurone layer. In wheat and maize, the outer differentiated layer of cells
of the endosperm; named thus because these cells are filled with aleurone
grains.
Allopolyploid. A polyploid having chromosome sets from different sources,
such as different species. A polyploid containing genetically different
chromosome sets; for example, from two or more species.
Amphidiploid. A plant possessing the sum of the somatic chromosome
numbers of two species.
Andromonoecious. A plant bearing bisexual or complete flowers instead of
strictly pistillate ones in addition to staminate flowers.
Aneuploid. An organism or cell having a chromosome number other than
an exact multiple of the nionoploid or basic number. Hyperploid
higher. Hypoploid = lower.
Anthesis. The period or act of flowering.
Apogamy. The development of a sporophyte from some other cell or cells
of the gametophyte (embryo sac) instead of from a gamete (egg).
Apomixis. The development of an individual from an unfertilized egg
without sexual fusion, whether the egg be normally haploid or abnor-
mally diploid through failure of reduction division.
Autogamous. Self-fertilizing.
Autopolyploid. A polyploid arising through the multiplication of the com-
plete genom complement of a species; e.g.j an autotetraploid has four
identical sets of chromosomes.
Awn. A bristle-shaped elongated appendage or extension to a glume,
akene, anther, etc.
Backcross. The cross of a hybrid to one of the parental types. The off-
spring of such a cross is referred to as the backcross generation.
Backcross method of breeding. A system of breeding carried out by several
generations of backcrossing and subsequent selection. The characters
of the recurrent parent are retained for the most part, and a few charac-
ters from the nonrecurrent parent are added.
Biometry. The application of statistical methods to the study of biological
problems.
Biotype. A population of individuals with identical genetic constitution.
A biotype may be homozygous or heterozygous.
401
402 METHODS OF PLANT BREEDING
Bivalent. A pair of synapsed or associated homologous chromosomes.
Bud-sport* A branch, flower, or fruit that differs genetically from the
remainder of the plant.
Bulk method of breeding. The growing of segregating generations of a
hybrid of self-pollinated crops in a bulk plot, with or without mass
selection, followed by individual plant selection in Fe or later
generations.
Caryopsis. A one-seeded dry fruit with the thin pericarp adherent to the
seed, as in most grasses.
Chaff. The floral parts of cereals, generally separated from the grain in
threshing or winnowing.
Character. One of the many details of structure, form, substance, or func-
tion that make up an individual organism. The Mendel ian characters
of genetics represent the end products of development, during which the
entire complex of genes interacts within itself and with the environment.
Chimera. A mixture of tissues of genetically different constitution in the
same part of an organism. It may result from mutation, irregular
mitosis, somatic crossing over, or artificial fusion (grafting) . There are
two main types, periclinial with parallel layers of genetically different
tissues and sectorial.
Chi-square (x 2 ) test. A statistical comparison of observed with theoretical
ratios.
Goodness of fit. Comparison of observed Mendelian ratio with a
theoretical.
Independence. A test for association between two series of variables.
Chromatids. Half chromosomes, resulting from longitudinal division, that
later became daughter chromosomes.
Chromosomes. Microscopically small, dark-staining bodies visible in the
nucleus of the cell at the time of nuclear division. The number in any
species is usually constant. They carry the genes, arranged in linear
order.
Class. A group that includes variates of similar magnitude.
Clon. All the individuals derived by vegetative propagation from a single
original individual.
Coefficient of variability. A measure of variability expressed in percentage.
Combining ability. The relative ability of a biotype to transmit desirable
performance to its crosses.
Complementary genes. Genes that interact to produce a new character.
Convergent improvement. A system of double backcrossing for the purpose
of improving each of two inbred lines without greatly modifying the
yield of their F\ cross,
Correlation coefficient. A statistical measure of relationship between
two or more series of variables.
Simple. The total correlation between two series of variables.
Partial. The correlation between two series of variables independent of
the accompanying variation due to other variables.
Multiple. A coefficient that measures the degree to which the dependent
variable is influenced by a series of other factors studied.
GLOSSARY 403
Coupling* The condition in linked inheritance in which an individual
heterozygous for two pairs of factors received the two dominant mem-
bers from one parent and the two recessives from the other parent;
y/e.0., A ABB X aabb.
Crossing over. The exchange of corresponding segments between the
chromatids of paired (homologous) chromosomes. It is a process
inferred genetically from new associations of linked factors and inferred
cytologically from new associations of parts of chromosomes, both of
which may be observed in heterozygotes. It results in an exchange of
factors and therefore in combinations of factors differing from those
that came in with the parents. The term genetic crossover may be
applied to these new gene combinations.
Cross-pollination. The pollination of a plant by pollen of a different
plant.
Deficiency, The absence, "deletion," or inactivatkm of a segment of a
chromosome.
Deletion. The absence of a segment of a chromosome involving one or more
genes,
Detassel. To remove the tassel, as in maize.
Dioecious. Having male and female flowers on different plants.
Diploid. An organism with two sets of chromosomes.
Disease garden. A special nursery for the study of reaction to specific
pathogens.
Dominant. A term applied to one member of an allelic pair of characters
-that has the quality of manifesting itself wholly or largely to the exclu-
v sion of the other member.
Double cross. A term used particularly in corn, where four inbred lines
are used as parents. The double cross is the F\ cross between two single
v' crosses.
Duplicate genes. Two separately inherited factors, either alone or together,
giving similar effects.
Duplication.- The occurrence of a segment more than once in the same
chromosome or genom.
Ear. A large, dense, or heavy spike or spike-like inflorescence, as the ear
of maize. Popularly applied also to the spike-like panicle of such
grasses as wheat, barley, timothy, and rye.
Emasculation. The act of removing the anthers from a flower.
Endosperm. The nutritive tissue formed within the embryo sac in seed
plants. It commonly arises following the fertilization of the two
primary endosperm nuclei of the embryo sac by one of the two male
"' sperms. In a diploid organism, the endosperm is triploid.
Epistasis. The suppression of a character dependent upon the action of
a gene or genes by a gene or genes not allelic to those suppressed.
Those characters suppressed are said to be hypostatic. Distinguished
from dominance that refers to the members of one allelic pair.
Euploid. An organism or cell having a chromosome number which is an
exact multiple of the monoploid or haploid number. Terms used for a
euploid series are haploid, diploid, triploid, tetraploid, etc.
404 METHODS OF PLANT BREEDING
Fi* The first filial generation. The first generation of a given mating,
F*. The second filial generation, produced by crossing inter se or by self-
pollinating the Fi.
Factor. The same as gene.
Fatuoids. Mutants occurring in cultures of cultivated oats and possessing
characters of A vena fat ua, wild oats.
Fertility. The ability to produce viable offspring.
Fertilization. The fusion of a male gamete (sperm) with a female gamete
(egg) and of their nuclei, without which their later development is
usually impossible.
Floret. A small flower, especially one of an inflorescence, as in grasses and
Compositae.
Foundation stock seed. Seed that has descended from a selection of
recorded origin, under the direct control of the original breeder, a
delegated representative, or of a state or federal experiment station.
Gamete. A mature male or female reproductive coll (sperm or egg).
Gene.- The hypothetical unit of inheritance located in the chromosome,
which by interaction with the other genes and the environment controls
the development of a character. Genes arc believed to be arranged
linearly in the chromosomes.
Genom. A complete set of chromosomes (hence of genes), inherited as a
unit from one parent.
" Genotype. The fundamental hereditary constitution of an organism.
Glabrous. Smooth, without hairs.
Glume. One of the two empty chaffy bracts at the base of each spikelet
in grasses.
Grain. Cereal seeds in bulk. Seed-like fruit of any cereal grass.
Haploid. An organism or cell having only one complete set of chromosomes.
Head. A dense, short cluster of sessile or nearly sessile flowers on a very
short axis or receptacle, as in red clover or sunflower.
Heteroploid. An organism characterized by a chromosome number other
than the true euploid number.
Heterosis. Hybrid vigor.
Heterozygous. The condition in which the homologous chromosomes of an
individual possess different genes of the same allelic series.
Homologous. Chromosomes occur in somatic cells in pairs that are similar
in size, shape, and supposedly in function, one being derived from the
male and one from the female parent. The two members of such a
pair are spoken of as homologous chromosomes.
Homozygous. Possessing identical genes with respect to any given pair or
series of alleles.
Hull. The term applied to include the lemma and palea when they remain
attached to the caryopsis after threshing.
Hybrid vigor. The phenomenon in which the cross of two stocks produces
hybrids that show increased vigor.
Hybrids. The progeny of a cross-fertilization of parents belonging to
different genotypes.
Hypostasis. See Epistasis.
GLOSSARY 405
Inbred Line. A relatively homozygous line produced by inbreeding and
selection.
Inbred-variety Cross. The F\ cross of an inbred line with a variety.
Inflorescence. The flowering part of a plant.
Interchange. An exchange of segments of nonhomologous chromosomes.
Interference. The property by which the occurrence of one crossover
reduces the chance of occurrence of another in its neighborhood.
Inversion. A rearrangement of a group of genes in a chromosome in such
a way that their order in the chromosome is re versed.
Keel. A central ridge resembling the keel of a boat, as in the glumes of
some grasses, etc. ; also, the inferior petal in the legume flowers.
Kernel. The inner portion of a seed within the integuments. Also the
t whole grain of a cereal.
Latin square. An experimental design for comparing treatments where the
number of replications is the same as the number of treatments and
each treatment occurs only once in each row and column. Especially
adapted for accurate comparisons with a small number of treatments.
Lattice designs. Designs developed for testing a large number of treat-
ments, in which the number of blocks exceeds the number of complete
replications.
Lethal gene. A gene that renders in viable an organism or a cell possessing
j it.
Linkage. Association of characters in inheritance, due to the fact that the
genes determining them are physically located in the same chromosomes.
Such a group of linked genes is called a linkage group.
Lodicule. A minute scale at the base of the ovary opposite the palea in
grasses, usually two in number, probably representing the reduced
perianth.
Mass selection. Selection for some desired character, or characters, where
progeny of the plants or heads selected are grown in bulk.
Mature-plant resistance. A term applied particularly to resistance to stem
rust in the stages from heading to maturity where this resistance is not
correlated with seedling reaction.
Mean. The arithmetic average.
Megaspore (macrospore). A spore having the property of giving rise to a
gametophyte (embryo sac) bearing only a female gamete. One of the
four cells produced by two meiotic divisions of the megaspore-
mother-cell (megasporoeyte).
Meiosis. The process by which the chromatin material becomes reduced
qualitatively and quantitatively to half the somatic number. It is
completed in the two divisions, meiotic mitoses, which precede the
formation of gametes in animals, or of spores in plants.
Microspore. One of the four cells produced by the two meiotic divisions
(mitoses) of the microspore-mother-cell (microsporocyte). A spore
having the property of giving rise to a gametophyte bearing only male
gametes.
Mitosis. The process by which the nucleus is divided into two daughter
nuclei.
406 METHODS OF PLANT BREEDING
Somatic mitosis. The process by which the daughter nuclei are identical,
quantitatively and qualitatively.
Meiotic mitoses. Two nuclear divisions that result in spores in higher
plants and in gametes in animals. Both divisions are necessary to
complete reduction.
Mode. The class of greatest frequency in a frequency distribution.
Modifier or modifying Gene. A gene that affects the expression of another
nonallclic gene.
Monoecious. With separate male and female flowers on the same plant.
Multiple alleles. A series of alleles in similar loci of homologous chromo-
somes of related races.
Multiple cross. A cross between more than two parental lines of different
origin.
Multiple -factor hypothesis. -The type of inheritance in which a character
is dependent on many different genes or factors.
Mutation. A sudden variation that is inherited. The term is used loosely
to include "point mutations" of a single gene and chromosomal changes.
Nonrecurrent parent. -IT sec I in backcrosses to refer to the original parent
not used in backcross generations.
Ovary. The swollen part of the pistil that contains the ovules.
Ovule. The macrosporangium of flowering plants, consisting of the nucellus
plus the integuments.
Pi, P^ etc. The first, second, etc., parental generation of a parent.
Palea. The upper of the two bracts immediately enclosing each floret in
grasses.
Panicle. A compound inflorescence with pedicel ed flowers, usually loose
and irregular, as in oats, rice, proso, etc.
Parthenogenesis. The development of a new individual from a germ cell
without fertilization.
Pedicel. A stalk on which an individual blossom is borne.
Pedigree method of breeding. A system of individual plant selection during
the segregating generations of a cross where the progeny plants usually
are separately spaced and the pedigree of particular selections is known.
Peduncle. The primary stalk supporting either an inflorescence or a solitary
flower. In grasses, the uppermost iriternode of the culm.
Pericarp. The mature or ripened ovary wall around the ovule.
Phenotype. The observed character of an individual without reference to
its genetic nature. Individuals of the same phenotype look alike but
may not breed alike.
Physiologic races. Biotypes or groups of biotypes within species that
behave more or less consistently in pathogenicity on certain differ-
ential varieties or host plants. Physiologic races sometimes are
differentiated on the basis of cultural or physiochemical characters.
Physiological resistance. A type of resistance due to physiological or
protoplasmic incompatibility between the host plant and the pathogen.
Pistil. That part of the flower consisting of ovary plus style and stigma.
Polyploid. An organism with more than two sets of a basic or rnonoploid
number of chromosomes, e.g., triploid, tetraploid, pentaploid, hexaploid,
heptaploid, octoploid, etc.
GLOSSARY 407
Proterandry. The maturing and functioning of stamens before pistils in
hermaphroditic flowers or in different flowers of the same plant in a
monoecious species.
Proterogyny. The reverse of proterandry.
Pubescent. Hairy, in a general sense; in special use, covered with short
soft hairs.
Pure line. A strain of organisms that is comparatively pure genetically
(homozygous) because of continued inbreeding or through other means.
Quadrivalent. Association of homologous chromosomes in groups of four.
Qualitative characters.- Characters that are qualitatively different, so that
separation is relatively easy.
Quantitative characters. Characters that show a continuous range in
variability, making separation into distinct classes difficult.
Randomized blocks. An experimental design in which the treatments are
arranged in random order within the blocks or replicates.
Recessive. -A term applied to one member of an allelic pair lacking
the ability to manifest itself wholly or in part when the other or domi-
nant member is present.
Recombination. The observed new combinations of characters different
from those combinations exhibited by the parents. Percentage of
recombination equals percentage of crossing over only when the genes
are relatively close together. Cytological crossing over refers to the
process; recombination or genetic crossing over refers to the observed
genetic result.
Reciprocal crosses.- Crosses where the parental plants or lines are used as
both male and female.
Recurrent parent. Used in backcrosses to refer to the parent to which the
first cross and backcrossed plants are crossed.
Reduction division; heterotypic division. Terms formerly applied to the
one of the mciotic mitoses at which a particular author thought reduc-
\ tion and segregation occurred.
"Registered seed. Seed of a variety or strain that is the multiplied progeny
of foundation stock seed and traces directly to it and that complies with
J certain standards of purity and quality.
Regression coefficient. A coefficient that gives the rate of change in one
variable (dependent variable) per unit rate of change in another
(independent variable) .
Replication. Repetition of treatments in experiments.
Repulsion. The condition in linked inheritance in which an individual
heterozygous for two pairs of linked factors received the dominant
member of one pair and the recessive member of the other pair from one
parent and the reverse condition from the other parent; e.g.,
A Abb X aaBB.
Rod row. A type of field plot approximately 1 rod long. Used particularly
\ with small grains where the seed is sown without definite spacing.
Roguing. The act of removing undesirable individuals from a varietal
mixture in the field by hand selection.
Seed. The mature ovule, consisting of the kernel and its integuments.
Also used for the seedlike fruits of cereals.
408 METHODS OF PLANT BREEDING
Segregation. The separation of the paternal from maternal chromosomes
at meiosis and the consequent separation of differences as observed
genetically in the offspring.
Self-fertilization. The union of the egg cell of one individual with a sperm
cell of the same individual.
Self -incompatibility. Some physiological hindrance to self-fertilization.
Sib mating. Crossing of siblings, two or more individuals of the same
parentage (brother-sister mating).
Single cross. A cross between two inbred lines.
Somatic. Referring to body tissues; having two sets of chromosomes, one
set normally coming from the female parent and one from the male, as
contrasted with germinal tissue that will give rise to germ cells.
Somatoplastic sterility. The collapse of fertilized ovules during the early
developmental stages.
Species. A group of individuals so much alike that it may reasonably be
assumed that they have arisen from a common ancestor.
Speltoid. Mutants occurring in cultures of common wheat, Triticum
vulgare, and possessing characters of T. spelta, spelt wheat.
Spike. A simple inflorescence with the flowers sessile or nearly so on a
more or less elongated common axis or rachis.
Spikelet. A small or secondary spike, especially in the inflorescence of
grasses.
Standard deviation. A measure of variability in terms of the units of
measurement. Frequently refers to the infinite population.
1 Standard error. Similar to standard deviation, except that it is calculated
from a sample.
Sterility. Inability to produce viable offspring,
Strain. A group within a variety that constantly differs in genetic factors
or a single genetic-factor difference from other strains of the same
variety.
Strain building. The improvement of cross-pollinated plants by any one of
several methods of selection.
Synapsis. The conjugation of homologous chromosomes.
Synthetic variety. A term used particularly with cross-pollinated plants to
refer to a variety produced by the combination of selected lines or
plants and subsequent normal pollination.
t test. A method for testing the significance of a difference.
Three-way cross. A cross between a single cross and an inbred line.
Top-cross. See Inbred-variety cross.
Transgressive segregation. The appearance in the F 2 (or later) generations
of individuals showing a more extreme development of a character than
either parent. Assumed to be due to cumulative and complementary
effects of genes contributed by the parents of the original hybrid.
Adequate testing of variation in the parents is required to establish its
occurrence.
Translocation. The change in position of a segment of a chromosome to
another part of the same chromosome or of a different chromosome.
GLOSSARY 409
Triploid. An organism whose cells contain three haploid (monoploid) sets
of chromosomes.
Trivalent. An association of three homologous chromosomes at meiosis.
Unit character. Term used for a character believed to be determined by
the alleles at a single-gene locus. The term is now largely abandoned.
Univalent. A chromosome unpaired at meiosis.
Variance. The square of the standard deviation or standard error.
Xenia. The immediate effect of pollen on the endosperm, due to the
phenomenon of double fertilization in the seed plants.
Zygote. The cell produced by the union of two cells (gametes) in reproduc-
tion; also, the individual developing from such a cell.
APPENDIX
TABLE I. TABLE OF /*
Degrees of
freedom
Probability (7*)
Degree? of
freedom
Probability (P)
.05
.01
.05
.01
1.
2
3
12 71
4 30
3 18
63,66
9 92
5 84
26
27
28
2 06
| 2 05
2 05
2 78
2 77
2.76
4
2 78
4 60
29 2 04
2.76
5
2 57
4 03
30
2 04
2.75
6
2 45
3 71
35
2 03
2.72
7
2 36
3 50
40
2.02
2.70
8
2 31
3 36
45
2 01
2.69
9
2 26
3 25
50
2 01
2 68
10
2 23
3 17
60
2 00
2.66
11
2 20
3 11
70
1 99
2 65
12
2 18
3 06
80
1.99
2.64
13
2.16
3 01
90
1 99
2 63
14
2 14
2 98
100
1.98
2.63
15
2 13
2 95
125
1 98
2 62
16
2 12
2 92
150
1 98
2 61
17
2.11
2.90
200
1 97
2.60
18
2 10
2 88
300
1 97
2.59
19
2 09
2 86
400
1 97
2 59
20
2 09
2 84
500
1 96
2.59
21
2 08
2 83
1000
1 96
2 . 58
22
2 07
2 82
OC
1.96
2.58
23
2 07
2 81
24
2 06
2 80
25
2 06
2 79
'"Abridged from Table IV of Fiwhei's "Statistical Methods tor ileseareh Workers,"
Oliver & Boyd, Edinburgh, and from Table 16 of Wallace and Siiedecor's "Correlation and
Machine Calculation," by kind penmasion of the authors and publishers.
411
412
METHODS OF PLANT BREEDING
-t to
to to
g
co eo
iO *H
CO tO
ei
&s
CO IO
CM (0
co to
O) 00
W
to"
2 8
oo to
1O W
* cn
co to
CO IO
CM -*
o
-* TH
'0 tO
SS
10 TH
^ CO
CO <#
CO
00 O
co cn
t b-
CM tO
^ CO
CD 00
S e?
'O
to"
CD cn
T-H cn
00 tO
00
iO CO
^
CO tO
CO IO
CM <#
CM
o
t eo
o 10
3 3
tO rH
tO CO
CO IO
00 b
CO O
CD ^
co cn
tO O
CM e
CO TH
CD en
co to
I-* CO
CM
to"
rH cn
00 tO
CO
10 CO
^ cn
co to
co 10
CM *
CM <*
O
eo ^i
LO CO
33
CD CO
>O CO
CO W
o eo
K cn
oo 10
CM b-
oo to
CO TH
b- ^
rH
CD cn
rH cn
00 tO
CO
TH
-t cn
co to
CO 10
CM
CM *
o
$ %
33
t- b-
o eo
ss
CM b
HH rH
ss
CD 00
CM b-
^ IS
t-
to"
C en
oo to
eo
to eo
TH
t cn
eo b-
eo 10
CO IO
CM <*
a
\^t
CM eo
.0 o
CM CO^
'"" *
28
00 W
to eo
00 tO
o cn
t>. to
6 eo*
rH
*t en
co* b-
CM IO
CO 00
CO W
CO tO
q o
co" id
0? 10
o
o
1 rH (0
S S
ft 3
-H
CD If
h- b
co cn
I- *
JT- TH
'f O
co cn
IO TH
O TH
CM tO
00 IO
rt2
^
,0"
CD cn
rH cn
00 tO
CO
to eo
^ cn
CO b-
CO IO
co 10
CM "*
"3
o oo
iO IO
CO b-
CM O
co 10
t- 00
^8
oo eo
00 00
co <n
00 O
CO
CO ^
oo to
* G
to"
CD cn
rH cn
oo to
eo
to eo
TH
*t cn
CO b*
eo 10
CO IO
CM **
^ 'Z
^
3S
10 to
CO tO
l^. CO
t^ cn
5
** TH
00 CO
^
CM 00
o co
fe G
o o
C"
CM
to"
CD en
oo to
eo
o eo
rH
^ cn
CO b
CO tO
CO IO
CM *
u o
X u,
(fi
C
3
-H 10
-t ^
co cn
cr to
s
CD 10
tO IO
K cn
oo eo
^ 2
2 eo
CO O
CD 00
C3
<t
CM
to"
CD cn
-H cn
oo to
eo
tO <#
t cn
co f
co to
CO IO
CM <
b T-H
S-.
CO
co cn
Tt O
CM TH
CO ^l
-t ^*
CD CO
CO 00
** IO
00 rH
9
to
CM CO
CD IO
CD b-
^ eo
O 00
CM <*
oo eo
CD cn
"^
to"
eo
rH
fe
-4 O
-s G
0)
t-.
bC
O
-f
>o eo
to"
CD cn
rH cn
".s
oo" to
00 CO
TH
HH b-
CD tO
CO b^
CM IO
O CO
co to'
co to
CM IO
eo id
CM O
co id
p>
o
2
-t tO
to"
-t <*
2 *
-f IO
00 t*
eo
CD CO
oo en
CO 00
"* cn
o eo
b;
CO )
00 b-
CM tO
co" id
r- TH
O TH
co id
a oJ
rH
co eo
-t* 00
5 3
CD CO
eo 10
CD ^
o to
r- en
co en
O b-
o *#
CO 10
CO b-
00
rH rH
s~l
rH
CD cn
00 eo
rH
rf cn
-t* b-
CO tO
eo 10
CO IO
/} "*"*
o
o
CM tO
-H IO
eo n*
00 CO
co ^*t
CD IO
bt
CO b-
O 00
co eo
CD tO
-t< eo
eo oo
co to
rH CO
H 0^
CD cn
oo b-
'0 *
rH
"t b-
co to
CO 10
eo w
^ -+-'
bC
cu
-f CO
00 00
eo eo
oo eo
o to
o to
00 IO
O CO
rH cn
00 rH
co b-
CD TH
co cn
00 10
rH CO
bC
-
rH cn
GO b-
CO
o *
rH
^
^ f
eo to
eo 10
CO 10
H
CO 00
CO CO
GO <<*
CM t-
00
>0
i ( rH
CO -^
b* co
38
co b
o"
CD cn
rH W
00 b-
eo
CO "*
rH
-*
rH
*t 00
CO tO
co to
co 10
J G
t- CO
CO CO
O *
co eo
oo 5
o en
00 IO
00 *
rH tO
CM CO
88
8S
CD CO
CM tO
1} o
-I rH
10"
CD cn
rH cn
00 N
CD *H
rH
't O
rH
-* 00
co b-
co to
CO IO
CO CO
CD en
rH CO
CD tO
CM *
i> en
00 rH
00 b-
to eo
eo 00
LO
10"
TH cn
00
CO IO
rH
t oo
co b*
co to
CO IO
*H
o
o <*
eo to
co eo
o eo
CO CO
CM 10
O b-
o en
CD IO
CO b-
r>. to
CD CO
co to
oo to
<* o
+H
02
10"
rH cn
CD 00
CO iO
o o
^ 00
co b-
co to
co to
0>
>0 IO
CM
tO IO
CM CO
CO S
CD cn
rH CO
CO IO
O TH
CM IO
rH 00
*f TH
00 O
co eo
CD ^
^
10"
CD en
-H cn
CD 00
CO IO
rH
>O TH
rH
-* cn
CO b-
co to
"*"
CO CO
rH O
CD t-
00 IO
CM *
S8
rH tO
CO 00
r e-
85
b- cn
O 10
00 8
to"
2S
og
co to
H
>O CO
^ en
-* oo
-* b-
CO tO
o en
o cn
83
*O TH
'O 00
CD 8
CD b"
t- CO
rt CO
rH cn
I- M3
s
co eo
CM
CD cn
" S
CO 00
>o eo
rH
10 o
rH
<* en
~* *
"* 00
S
io 3
CO CO
rH rH
8
CO CO
CD S
CD IO
'O CO
CM tO
CO CO
CM tO
rH IO
^ S"
oo oo
rH cn
O ^f
rH CO
c
CO tO
rH
o eo
rH
' rH
tO TH
rH
'O O
9
C
-
CO
,
.
CO
-
00
Oi
APPENDIX
413
ST*
cn
s
88
CM IH
eo g
fe
rH IO
O b
s
01 b-
O5 IO
285
S8S
rH (O
00 CO
CO
CN W
CM CO
01 eo
CM CO
04 CO
CN 04
Ol 04
rH 04
rH 04
rH 04
rH O4
TH* 04
Jo en
^8
rH CO
co eo
CM 00
CM TH
rf
g
o
05 5
co en
05 10
O5 IO
>O *
00 <#
00 CO
&
CM CO
CM CO
04 eo
oi' co'
01 eo
O4 04
O4 O4
rH 04
rH 04
rH 04
rH 04
rH O4
ft
co to
10 en
CM tO
*# to
04 TH
CO ^
Ol O4
28
O C4
O CO
05 O
a tfi
O5 IO
00 5
00 ^
CN CO
CM CO
O4 CO
CM CO
CM CO
04 C4
O4 04
rH CN
rH 04
rH 04
rH O4
rH 04
o3
o o
iO O
o to
CO *
CO b
04 04
05 TH
rH TH
Ol b-
t- to
O CO
01 tO
C b-
00 00
Oi O
-t O
OS tO
S
l^. b"
00 ^
CM *
CM CO
CM CO
O4 CO
CM eo
O4 04
O4 04
CM 04
rH 04
rH 04
rH 04
rH 04
~3
CO O
rP i
CD en
CO <*
00 O
O4 CO
CM TH
2
00
3?
TH
b-
co eo
05 tO
04 tO
O5 IO
/
cu
rj
CN *#
CM CO
CM eo
CM CO
Ol CO
04 CO
OJ 04
O4 04
04 O4
rH O4
rH 04
rH 04
CO TH
CM' *
04 CO
oi eo
OJ b-
CO CO
O4 CO*
O4 O4
CM* CO
*
CM eo'
CO tO
C4 04
DO tO
O 00
04 04*
O C^
oi w
O4 04
CO CO
O3 tO
rH' 04
CO 00
O5 IO
I
CO TH
~f <0
Cl 04
Ol TH
O 00
O b-
05 tO
O5 tO
'3
CM "#
CM CO
04 eo
04 CO
ot eo
Ol CO
CM CO
04 04
Ol 04
04 04
rH 04
rH C4
5
8
i o en
00 TH
co eo
ic o
04 04
^ S
12
^ en
t- *p
O CO
O b-
04
b-
T.J
CM *
CM CO
04 CO
O4 CO
CM CO
04 CO
O4 CO
oj eo
04 04
CM
O4 04
O4 O4
-3
K w
CN <*
rH C4
CD O
04 *
O 00
iq t-;
oi co*
04 cn
O4 CO
10 eo
CO ^J
oi co
CT< cn
O4 CO*
-f 00
04 TH
Ol CO*
Ol *>
OJ CO
28
oi n
rH C4
rH q
Ol 04
oo to
O 00
04 0*
c o
O 00
O4 04
of
CM" ^
CM 04
00 10
CM" ^
cgS
s$T
t en
C 00
04 CO
cToo~
co en
04 CO
CO b
"t to
OJ CO
04 CO*
CO IO
c-i eo
01 CO*
co to
CO CO
04 CO
co 5
04 CO
00 IO
O4 04
oi eo
co eo
04 CO*
co to
CM TH
oi eo
"8 5
O4 CO
rH" O
OJ CO
01 CO
iC O
ri O
04 CO
04 TH
CM CO*
CJ ^
rn en
oi 04
CO tO~
rH O
O4 CO
00
00
oi O4
io cn
rH en
04 O4*
O t-
o
<u
s
r/T
oo to
CM i*
oi *
oj eo
O4 CO
04 CO
CM eo
OJ CO
04 CO
Ol CO
OJ CO
OJ
Ol CO
a>
(M
O5 b-;
CM <#
3
oi *
O5 tO
CO TH
01 *
o to
co q
O4 CO
CO O
oi eo'
00 b-
t to
oi eo*
OJ 10
"f 10
oi co
00 10
CO ^f
04 CO
-t b
CO CO
04 CO
CO CO
04 CO
00 CO
04 04
04" eo
>0 b-
04 TH
O4 CO
1
"? 00
O5 b-
CM
04 tO
00 <#
04 O4
I- 04
04* **"
CO 04
CO
04 "**
;J
rH CO
O4 CO*
01 eo"
Tf O
<N* CO
CO <*
Ol CO
-H tO
CO CO
01 eo"
co co
oi co
00 ^H
01 04
oi eo"
u
cu
l> 10
O5 00
CM "*
O ^
oo 10
CM* ^
8
01 ""**
b, o
CO TH
oi -*
s
O4 CO
>c o
ic oo
O4 CO
??
O4 CO
>c en
't IO
oj eo
oi co"
GO CO
CO <*
04 CO
iC b-
CC CO
01 CO
Ol t-4
CO CO
oi co
m
<A
CM 10
o q
CO* *
o eo
o> q
00 CO
04 ^
04 cn
04 *
o eo
co q
01 *
p5 en
01 CO
f CO
CM CO*
C no
>o to
oi co
CO O
Hi tO
04' co
CO 04
~V 10
04 CO
C IO
oi eo
r^ O
O4 CO
o
cu
o q
CO 10
O5 b^
CM* "*
>o o
00 IO
04 ^
t- O
1^ CO
o" *'
o **
r- TH
O4 *
co q
O4 *#
CM" co'
>C b
04 CO*
CM* CO*
01 co"
c to
oi co'
04 TH
*t 10
oi co
t TH
CO 10
rH 00
O 00
co tj*
04 IO
O4 "J
00 ^
O4 -^
t^ 00
Ol d
O "*
t^ TH
01 -"di
CO CO
co q
O4 T}i
oj eo
co q
O4 CO
00 IO
>c oo
CM* CO
Ol CO
04 TH
>O t;
oi co
O5 IO
Tf! (^
oi co
o
OJ
CM en
04 eo
O5 b-
O 04
O 00
Si 2
.0 to
00 <#
C5 04
t>. CO
f O
C O
CO TH
CO O
co en
b-
co oo
iC 00
Is
CO ^<
CO tO
O 04
O4 CO
rH tO
01 to
o oo
s
o to
O3 10
00 *
00 CO
(-- IO
t^ TH
rH O
J>> TH
00 "*
CO
aS
f/)
oo en
85
CO TH
04 *#
% 8
^ s
co en
o oo
O b
CO b-
O5 tO
CO 00
05 IO
o o
05 10
3
t b
QO eo
o
rH IO
IO
38
% cn
*
-H tO
CO IO
83
04 04
O 00
O4 TH
co cn
rH O
Or, ,H
rH O
o ^*
TH cn
O 00
o
T3
oo to
co to
CO IO
CO IO
CO IO
CO IO
CO IO
CO IO
co 10
CO IO
co ^
CO ^l
28
** b-'
00 O
05 04
CO* b-"
00 CO
oq en
co" to
GO b
co to
"t TH
l> 10
co to
oo to
CO CO
co to"
CO CO
CO O4
co % to
O5 TH
CO* tO
'"q O
co" to"
01 eo
10 cn
co ui
O5 IO
""t n .
CO IO
N. 00
CO 10
G
(_l
CO "*
05 O
-f 6
"f IO
GO tO
to eo
t>- CO
co q
CO 00
3 &
~t< CO
t CO
"* 06
^ 00
00 00
CO TH
10 O
CO TH
OJ Cf4
CO O
a>
"c
rH
Y-H
rH
Cl
ro
3
rH
CO
T t
rH
00
05
s
O4
a
OJ
414
METHODS OF PLANT BREEDING
| SS
co to
S
^ S
K O
CO 8
CO
O O
1 *HM
rH Od
rH Cd
rH Cd
rH Cd
rH Od
rH (N
rH Cd
O
oo eo
fcS
-t CO
CM O)
O IO
00 Cd
tO tO
CO O
** CO
CO O
rH Cd
rH Cd
rH Cd
rH Cd
rH Cd
rH Cd
o
rH |>
" CO
CD to-
't CO
CM O
rH tO
Ol CO
oo o
CO t-
rHCd
rH Cd
rn cd
rH'c4
rH Cd'
rH Cd
rH Cd"
rH Cd
00 ^**
CJ to-
oo eo
00 CO
t^. en
CO IO
1> Cd
cj co
l> rH
!> TH
01 eo
CO H
rH
rH Cd
rH Cd
rH 01
rH (N
rH 09
TH <N
rH Cd
rH Cd
rH Od
CD
t- to
oo ^Jj
rH Od*
HH TH
00 O
&S
O Cd
oo eo
2
co to
rH Cd
rH Cd*
CO O>
rH Cd'
CM tO
O TH
rH Od
PH
r>J
o
rH eo
Ol IO
00 3
S3
"t O
oo ^<
01 tO
oo eo
O eo
oo eo
00 O
t*. eo
r^ c-
1- Cd
CO -^
t^ Cd
rH Cd
TH Cd
rH Cd
rH Cd
rH Cd
O
-v d
00 00
5s 10
rH CO
Oi IO
a o
oo -^
I- IO
oo -i
IO TH
CO ^
oo eo
rH IO
oo eo
O Cd
oo eo
R S
S T=
x-v
rH Cd
rH Cd
* n
d
o
CO
00 to-
Ol to
22
~t 00
CM *
Ol IO
Oi IO
00 t-
00 *
00 <#
rH Cd
tO TH
00 *
TH Cd
oo eo
rH* Cd
5 "*
->
-' su
C
a!
Cl
CO iO
o t-
01 Od
g
Cl Cd
oo to
Ol tO
is
rH Cd
S3
C*i IO
O5 IO
rH Cd*
05 IO
rH Cd*
3
rH Od
%$
rH Cd
o
rH
' CD
* ft
Cw
1
o
Cl
b eo
o oq
CM Cd
-t CO
l>
CM *
O to-
Ol Cd"
c i>
01 Cd
rn" Cd
t>- eo
CO O
Ol IO
rH Cd
CO IO
01 10
rn" Od
t, ^
5 O
bC
o
co
CO *
rH <*
cj cd
O
rH 00
Cl Cd
o oq
q oq
'0 t-
q t-^
CJ Cd'
CO <**
q i>
Cl" Cd'
q to^
CJ Cd*
o oo
tO
OJ Cd
Ol to
01 to
rH Od'
>> 4X
*
D
c
-c
00 Cd
cj eo
"t to-
co eo
CM Cd"
01 Cd*
S oq
Cl Od
00 CO
o oq
CJ Cd'
CO O
o oq
Cl Cd'
q l>
SS
Oj' Od
a -s
5 ^>
*
Q
0)
2
CO Cd
CM TH
CM" eo"
to-
ci q
CM eo
oo eo
rH O
old
SS
T3 CO
rH O>
CJ* Cd'
01 o
rH Cft
ci c4
O to-
rn oq
CJ Cd"
ci Od"
^ S.
3 >.
a-
be
^
co eo
CJ TH
CM eo'
CJ CO
CM q
cj eo'
C4 CO
00 Cd
01 eo'
CO 00
OJ* Cd'
>0 10
01 Cd*
"t Cd
CM Cd
01 O
CJ Od"
CP
tJ
o
o to
CO Cd
00 TH
CD to-
CJ rH
HM eo
Cl rH
CM <n
01
g
ss
S
co eo
rH 9>
s S
C
Cl CO
cj eo
01 CO
Ol CO
oj eo
OJ CO
OJ CO
Ol CO
CM Od
' "c
H ..H
o o
co eo
co eo
o o
CO Cd
Cl Cd
t- to-
ci TH
o *#
Ol TH
CM TH
01 00
OJ O
<N
D bC
CM eo
CM CO
cj eo
Cl CO
01 eo
CJ eo
ci eo
ci eo
ci eo
5 "H
00
S?J5
cj eo"
00 TH
co ^
oi eo
co eo
'f Ct
CO CO
cj eo*
CM eo
o to
co cd
CM CO*
en eo
OJ Cd
CJ CO*
00 O
01 Cd
OJ CO
04 TH
oi eo*
t>- a>
>o ^
5 S
rH &
Ol Cd
r- at
co yj
CO
*t o
. o
H
*-
cj eo"
CJ CO
cj eo
CJ CO*
01' eo*
oj'eo"
ci eo
oi eo
ci eo
< 0)
iO to-
CO TH
iO C-
S 5
a eo
f tO
t- o>
CO tO
H? eo
CO O
tj
I
i j_,
' o
CM eo
cj eo
co at
ci eo
Cl
cj eo
CO 00
cj eo
iO 00
cj eo
tO to-
CJ CO
~~cc~to
to to-
oj eo
Hn CO
to t-
cj eo
~co~0
to to-
cc
CM eo
cj eo
CM CO
cj eo
CM eo
Cl eo
ol eo
cj eo
cj eo
2
oo eo
06 cd
00 (N
CO 00
tot TH
CO TH
t- TH
i-^. o
o ^
Ol Cd
CO O
crj
Cl ^
CM ^
OJ *
CM *
OJ ^
04 -s*i
CJ -^
CM ^
CJ ^
K"
CO
00
CO tO
o t-.
to-
01 CO
oj 5
00 ^
g
>O to-
CO ^
Ol 10
CJ TH
Oi 10
3 to2
sjs S
O TH
oo to-
co 10
b. eo
CO IO
C?^
H^ IO
CO #
83
01 en
CO CO
CO 10
CO IO
CO IO
CO 10
co 10
CO IO
CO IO
CO IO
CO 10
rH
CO 9*
oo oo
CM 00
CO Cd
04 CO
"f fr-
ci to"
OJ Cd
d fr-
S
c *#
04 to
oo o
i> to
<* to-
S
CM
CO
CM
a
to
CM
01
CM
00
Ol
04
o
CO
APPENDIX
415
Oi tO
tO en
b* H
<o en
Ift fr-
iO 00
ee H
10 oo
rH ^
10 00
Ol 00
T^ t-
oo to
Tf O
ee 64
^ l>
1C O
* t*
^ co
"# tO
rH ^(
^ to
a> o
co S
f
J3
r-< 00
CO en
OS ^K
10 en
CO O
10 en
Tf <O
O 00
CO i*
HO 00
rH O
to co
^S
oo to
t> CO
CO ^
CO O
rH CO
-t o
rH TH
rH TH
rH ^
rH T-l
rH *S
rH <H
rH rH
rH TH
rH TH
rH T-l
J3
3
D.
Tti en
CD O
rH 00
CO en
Ol ^
o
^ O
>o o>
IO 00
to oo
*t IO
tO 00
01 en
iO 00
rH O
iO 00
Sg
00 <O
Tt t*
CO *4
<* e
T(1 00
^ to
-T!
pj
iH C*
c
b- 00
CO O
rf* *
CO O
w o
CO O
CO CJ>
O5 ^
10 0>
r- i-i
iO O>
CO 00
10 CO
-# to
>0 00
CO -^
iO 00
01 04
tO 00
CO
o t-
^2
O
*d
rH O4
rH 04
rH 09
rH rHI
rH rH
rH ,H
rH *H
rH T-l
rH T-l
rH TH
T-H T-l
rH T-l
rJ
Ol O4
CO TH
t- 00
CO O
>o ^
CO O
CO O
CO 5
rH C-
co cr
ss
oo 04
>o en
I- O
to en
CO' 00
tO 00
to to
tO 00
^s
S?
r
t O
b- 04
1^ TH
01 ca
CO T-l
C0
co to
CO O
co O
s
O4 00
co en
S
o ^
co en
00 O
to en
CO t-
>O 00
c;
"CD no ~
I>- 04
~ "+T TH ~
!> 04
o7~i>"
r^ *-<
"^T^T""
r>- rH
~cT <H
CO iH
"~ 00^00~
CO O
~I8~
~! s"
~S 8
co~cT
co o
rH~ &~
CD O>
oT""co
o <n
2
1-4 04
r-< en
CD
01 *
00 CO
'00 eo
00 (O
1- 0<l
CO N
r- cxi
Z3
C5
Cl IO
f-^ t-l
rH CO
t- TH
O T-l
t^ <H
01 o
CD H
r>- to
CO O
iO CO
CO O
O-
TJ
T-( 04
rH 04
iH W
rH
rH 04
rH 04
rH 04
TH 04
tO 04
00 <
Tf CO
oo eo
(M IO
QO eo
O N
oo eo
O5 cr>
i-
00 IO
i en
CO *
t^. Ol
'O 04
>. 04
-t O
t^- C4
^ 00
t T*l
01 to
IN. -H
04
I> rH
">>
rH 04
T-T~TH~
as 10
rH 0)
00 *
~^ "eo"
00 ^
~~7~~o ~~
00 ^
^~t-
oo eo
(N~W~~
00 CO
ii Ofl
00 CO
__^_-
oo eo
Q^Q^
t- O4
" 00"^""
t- 04
~_____
J^ 04
LCTO
t^. 04
of
if
o
I~~e4~^
05 10
"o ob~
O5 O
cr o
^j~^i ~
o; to
O~CT>"
Oi *
~~o~to
at. ^t
oo~^i~~
00 ff
__^,___
00 T
""C 1 "o"
00 *
o~ o~
oo co
P0~"
oo eo
T-H 04
00 CO
8
2
rH 04
<j
CM O
O t>
c to
00 N
0> 9
co er
Ci
>o to
a 10
^ ^
a> w
Ol N
C- 10
si
O CO
^
o to
a> *
00 CO
oo *
5
8
r^~o^
q co
5=r to~
o t>
~~co cT"
O b;
ci~oT ~
o to
to
c^ ""* ~
a> <o
~oo~~ci ~
05 tO
___^___.-_.
OS tO
"co^oo
OS 10
"~S3 T"
OS 10
coTeo ~~
O5 10
~w b
O5 IO
a*
<t>
bC
rH 04
2$
00 04
O 00
CO CO
t-
in IO
O f
o t-
c-j o
o t-
rH CO
o to
tO
to
a ^
01 tO
00 04
Oi to
r^ en
Ol IO
to to
Oi 10
01 04
CM 04
M e*
CN M
O4 N
04 N
01 CXI
01 e*
rH 04
rH 04
rH 04
rH 04
S
oS
"tf 5*
rH Cn
oi 04*
<N en
rH 09
Ol 04
2?
C Oil
C: Od
O 00
(N Ofl
r- o
O 00
<>i CA
co fr.
O l>
<N N
id IO
O fr- %
01 en*
^ eo
O C^
OJ 04
CO -r*
o t>
01 O4
<N
t-.
Ol 04
o to
o to
oi 04"
ss
5
1
ss
rH Cf
rH O
rH O>
rH 00
S 5
O 00
O CO
oS
t
3S
O
Ol CO
(M 04
<M exi
<M CXI
cj exi
<N <SI
OJ 04
01 04
CM 04
01 04
01 O4
01 04
-d
tO 04
01 i*
co oo
01
<N O
28
00 CT>
rH CT
N- tO
T-H en
CO ^
r-i Cft
"t 04
rH 0>
t
rn 0>
CO CO
rH 00
rH CO
O O4
rH CO
H
CM CO
Ol CO
<M eo
<>j eo
(N Od
CJ C*
01 04
01 O4
Ol O4
01 04
01 O4
Ol 04
of
01 D
CO O4
O T-l
CO 04
oo co
^4 T-l
10
C4 tH
10
-f O
CM i-H
CO l>
01 O
Ol IO
01 O
rH ^
Ol O
O 04
01 O
00 00
rH en
t- IO
rn en
O
^J
<M eo
01 eo
co eo
<N eo
(M CO
cq eo
OJ CO
01 CO
Ol CO
01 eo
01 04
Ol 04
OJ
O 04
-t ^J
oi eo
00 00
CO CO
<M eo
CO 10
co_ eo
c4 eo"
tO CH
co eo
<N CO
t Oi
CO
oi eo
04 tO
CO
oi eo'
^H <()<
co en
oi eo"
O O4
CO 04
oi co"
s
01 co'
Oi 00
01 T-l
oi co*
b> IO
01 T-l
oi eo
>O 04
Ol TH
oi co
"e5
o
'*3
rH (O
>O (O
Ol rH
<*
^
CO ^
Tf U9
tO rH
^5
c^ to
f ^
01 **
^ *x
%%
O T-l
*K *
oo e
CO CO
t-. ^
co eo
co eo
01 eo
CM CO
<N eo
<M eo
w eo
CM CO
01 eo
Ol CO
01 CO
01 eo
M eo
OQ
*- t-
CO 04
Ol CO
10 eo
co en
oi co*
co en
CD CO
<M" co
(N IO
CD CO
ci eo
r-4 CO
CO 00
ci co'
OJ O
>q co
ci co"
00 00
iO fr^
01 CO
r- to
to t-
oi co"
CO -^
q i>
oi co
CO N
tO fc-
oi co"
"t CO
10 S
oi eo"
Ol IO
iq <o
oi eo'
^05
""-(
o
o to
01 *
oo 04
QO *
O 00
oo eo
>o *
00 CO
^ rH
00 CO
CO O>
00 04
01 to
00 04
rH *
00 04
O O4
00 04
o> o
i-* en
oo to
l> T^
co eo
l> TH
cu
T3
(M <#
c
o -^
CO OS
00 Cn
<M 04
CO 10
CM e*
IO T-l
CN <N
CO 00
(N Y-l
0^ IO
<N t-
S3
O O
01 -l
OS 00
rH O
oo to
^H O
^s
>O 00
rH en
CO
r)
CO IO
CO IO
CO HO
CO IO
CO IO
co 10
CO IO
CO tO
CO IO
CO 10
CO IO
CO *
2
2
CO ^
rH TJ*
38
o o
rH CO
oo *-i
O CO
N. t
O <M
CO *
o en
O -l
04
"f en
O **
CO t
THI
01 04
*4
00
TJ
^ l>
Tjl fr
^* fr-
^ t
Tt< t-
^ b-
f fr-
TH |>
"*l t>
rH C-
*# fr-
* fr
416
METHODS OF PLANT BREEDING
8 co to
10 eo
eo 10
c^5
S3
iO t-
01 eo
O) CO
CM eo
OS 00
TH 04
W
rH rH
00 i-l
O *H
88
E
0)
o SS
IN. tO
CO IO
o e
CO 10
33
S3
iO If
01 eo
01 eo
oj eo
CO ^
TH 04
co en
TH rH
rH \[
a
1C r " 1 **
1-1 1-1
!-< 1-1
TH iH
TH "H
TH t-4
TH rH
rH rH
TH rH
rH rH
a
Gi
, ^ ^
^ ^
o
TP
00 If
CO IO
t -
CO O
rH tO
CO ^*
S3
co cn
01 eo
O4 O4
01 eo
28
Is. 11
13
(M TH TH
TH 1-|
d
_ CO rH
% *+.*".
>o o>
^ D
01 IO
"t W
o> o>
CO <*
^ H
04 00
00 04
CO 00
HH tO
O
X3
2 TH rH
rH 1-1
rn rH
TH rH
"5
as o
10 -f t-
* TH* 1H
t- *
"^ ^
r4 i-H
iO O
t ^
r4 <H
01 *
"*. *?
t-H 1-J
os en
CO US
C0
o eo
co 10
TH* TH
01 If
CO ^J
rH rH
gs
rH rH
00 rH
01 ^
0)
rC5
p,
\*t
Tfl ^)
o iO 00
>o ,_| ,4
CO N
'O 00
i-H 00
o t>
00 CO
Tft f
00
^ <,
* IO
T}J UJ
Ol 04
^ O
00 t-
CO IO
CO *#
CO 10
'O 64
CO IO
O
-->
o
s o3
? %
o ^ o
^ rH' rH
CD 00
CO
r-I r4
-t ^t
q 00
T-H 4
rH O>
>o t-
i-i -i
o to
"t ^
TH 4
l>- M
't ^
TH T-i
-. s
rH i4
01 #
~& o
rH rH
TH rH
"* *
TH rH
o en
H^ W
.2
2
*
"o
s: ^
CO Q
co 5
co ,_< ej
(N 00
O tf
o *
CO 0>
t^ en
>O 00
IO
ic oq
'f CO
to co
01 cn
tf
OS Tt
t_ t-
O rH
"t* t-
co cn
t to
T3
fl
c -
'**!
-
O *-' 3
oo o>
44 CC O
CM rJ 4
t^ t-
o o
Y-H* C9
o eo
o q
,-J Qjj
CO 00
co en
o ^
e o
OS -
^ i
t- 00
iO 00
't ^
>C CO
CO rH
*o eo
01 en
tO t-
I>v
-D
o o 4
-
U?
eo co
t- t-<
01 rH C4
01 IO
K -H
T-H ci
O r^
t^ rH
rH C<l
oo to
CO
1 CNJ
o eo
co
r-H OJ
38
TH ci
01 t-
co cn
rH TH
co en
oo en
'O 00
jr^ t
tc oo
EC
O
: i
^ C 3
1 O O
to I oo eo
C5 eo
J^ OQ
rH M
1^ -^
> N
rH C4
o en
I- r^
r-4 04
01 10
(> 1-1
i-l c4
jqa
TH e4
o; cn
TH 04
s
rH 04*
>O rH
co- q
rn" 04
-t cn
cc_ en
6
<!
^ gi
H g M
'O If
^ 00 W
*f IO
oq n
(M W
oq eo
Ci O
t- M
K eo
r- N
ID O
b-
t If
t^ *H
Ol C4
t^. rH
o cn
** 9
a> b-
CO O
m
co
OJ
& % <3
C4
04
fl*
!>i
!/) - ' i
01 Oi "*
.H _ N
cr. 10
00 *
rH 04
00 i-H
00 "*
T-H W
o to
00 CO
T-H
co eo
oo eo
TH c4
01 o
oq eo
TH ei
O 00
00 04
oo eo
r- C4
TH 04
CO O
ts ; N .
TH 04*
tc oo
r- i"
TH' 04
S
0>
5 a 1
3 > **
HH ^1
T-l OS IO
T- 1 TH 04
CO H
CJi 10
T-t 00
Qi ^
r^ Oi
oo eo
00 ^
TH
CO O
00 *
TH' o4
ic e-
oo eo
TH c4
CO *
oo eo
TH' 04
00 04
TH 04*
o to
oq 04
rH 04
OS <*
r>. e>
TH
o
0?
c3
3 HH o
^ ^ &
00 rH
O OS tO
K en
o 10
>t IO
Ci 10
Ol i-l
OS
5
OS <*
00 <*
t^ 1-1
oq ^
tO If
oq eo
SS5
CO 04
oo e*
W
* % g
w
en
C4
N
2S 1
01 o
-, f
a rvi 04
qo
O5 <*
C- CO
i> en
a 10
iO O
a to
HH CO
q 10
gs
O *0
q ^j
a eo
oq ^
00 rH
00 *
O
r 1
cu
be -
04
C4
A
r^
H C
00 Cn
00 ^ ^
fef:
t*
O l>
co en
o o
oS
O
O <O
?8
CO IO
0> W
to eo
a 10
^S
H
3 '^
. +j
00 01 en
04 CM
Ol [
Ol W
04
01 04
TH 04
TH 04
rH 04
TH 04
to"
T3
H c
<D
. C3
3 H
tO eo
^ o>
oi 04
Tf< f-
rH OJ
tN
Ol l>
^ *.
01
W
TH 00
04 '
oo en
If
01 e4
t- (O
q if
oi C4
to eo
O If
04 04
^o en
Ol 04*
01 tO
q co
01 04
ss
oi
0)
rH
; a
"f O
N - ^
CO C-
CN
r-i ^
01 o
28
JN *e
rH 0)
cr oi
TH en
-t O
TH cn
04 IO
rH CO
O C4
rH CO
s
o 10
oi eo
a eo
01 eo
01 d
Ol 04
01 04
01 04
01 04
01 04
01 04
3 tH
CD
CO T-l
co w
esj w
>o en
CO W
w eo'
CO IO
CO N
<N eo
o o
CO CS|
oi eo
01 t-
Ol i-<
oi eo
r^ -^
O4 H
oi eo
CO i-l
01 i-J
oj eo'
co <o
oi q
oi eo
04 ^
01 o
oi eo
04 CO
.a
cfl
t
<L>
j3
13
^ eo
iO U)
"^ cs eo
04 eo'
00 <O
~f IO
01 eo*
CO 1-
"* IO
01 eo'
t t-
"t "^
oi eo'
CO *
^ ^
oi eo*
rH rH
*f *
oi eo"
o <o
co eo
oi eo'
00 ^
co eo
oi eo'
l> M
co eo
01 eo*
**IU
o
iO O
CO ^ ^
t 00
I- O
01 *
t* G
O 00
t^ <n
00 <#
CD en
t- Y"t
CD
ss
01 CO
CD CO
S 00
O 00
CO If
8
05 (M
OJ *
Ol ^K
01 eo
01 eo
04 eo
M eo
O4 CO
oi eo
O4 CO
o
C
^ IO
d M . J
CO M
tH 0>
S3
a
00
r- oo
t-
g
t rH
O h*
04 O
o o
C4
to
Os O
Oi tO
Ctt
p
Oi ^
^ *
00 i-l
OS O
CO (0
ci o
^ o
a en
Ol <*
Oi 00
TH T-l
a> co
Oi
00 f
CD
00 t
>O CO
oo to
* *
oo to
T3
CO t^
CO t-
CO (0
co to
CO
CO (O
CO (O
CO O
co to
co to
H%
s
iO
CO
o
l>
i
o
o
iO
01
8
8
nf
1
8
i-i
&.
eu
PH
)*
APPENDIX
TABLE III. TABLE OF x 2 *
417
Degrees of
freedom
Probability (P)
.99
.95
.50
.20
.10
.05
.02
,01
1
0.0002
0.004
0.46
1 64
2.71
3.84
5.41
6.64
2
020
103
1.39
3.22
4.6(
5.99
7.82
9.21
3
0.115
0.35
2.37
4.64
6.25
7.82
9.84
11 34
4
0.30
0.71
3.3b
5,91
7.78
9.49
11.67
13.28
5
0.55
1.14
4.35
7.29
9.24
11.07
13.39
15.09
6
87
1.64
5.35
8.56
10.64
12 59
15.03
16.81
7
1 24
2.17
6.35
9.80
12.02
14.07
16.62
18 48
8
1.65
2.73
7.34
11.03
13 36
15.51
18 17
20.09
9
2 09
3.32
8.34
12.24
14.68
16 92
19.68
21 . 67
10
2.56
3.94
9.34
13.44
15.99
18.31
21.16
23.21
11
3.05
4.58
10.34
14.63
17.28
19 68
22 62
24.72
12
3.57
5.23
11.34
15.81
18 55
21.03
24.05
26.22
13
4 11
5 89
12 34
16 98
19 81
22 36
25 47
27 69
14
4.66
6.57
13 34
18.15
21 06
23 68
26 87
29 14
15
5 23
7.26
14.34
19 31
22.31
25 00
28 26
30.58
16
5 81
7.96
15.34
20.46
23.54
26.30
29.63
32.00
17
6.41
8.67
16.34
21.62
24.77
27.59
31 00
33.41
18
7 02
9 39
17 34
22.76
25.99
28.87
32 35
34.80
19
7.63
10 12
18.34
23.90
27.20
30.14
33.69
36.19
20
8.26
10 85
19.34
25.04
28.41
31.41
35.02
37.57
21
8.90
11 59
20.34
26.17
29.62
32.67
36.34
38.93
22
9.54
12 34
21.34
27.30
30.81
33.92
37.66
40 29
23
10 20
13 09
22.34
28.43
32 01
35.17
38.97
41.64
24
10.86
13.85
23.34
29.55
33.20
36.42
40.27
42.98
25
11 52
14.61
24.34
30.68
34.38
37.65
41 57
44.31
26
12.20
15.38
25.34
31.80
35.56
38.88
42,86
45.64
27
12.88
16 15
26.34
32.91
36.74
40.11
44.14
46 96
28
13.56
16 93
27 34
34.03
37.92
41.34
45.42
48.28
29
14.26
17.71
28.34
35.14
39.09
42.56
46.69
49.59
30
14.95
18.49
29.34
36.25
40.26
43.77
47.96
50.89
For larger values of n, the expression -\/2x 2 ~~ "\/'2n 1 may be used as a normal deviate
with unit variance.
* Abridged from Table III of Fisher's "Statistical Methods for Research Workers,"
Oliver & Boyd, Edinburgh, by kind permission of the author and publishers.
418 METHODS OF PLANT BREEDING
TABLE TV. TABLE OF r, FOR VALUES OF z FROM TO 3*
K
.01
.02
i
.03
.04
.05
.06
.07
.08
.09
.10
0100
0200
. 0300
.0400
. 0500
. 0599
. 0699
. 0798
. 0898
0997
1
. 1090
.1194
.1293
.1391
.1489
.1586
.1684
.1781
.1877
1974
f) 2
2070
2105
2260
. 2355
2449
2543
2636
2729
.2821
2913
. 3
3004
. 3095
3185
.3275
3364
3152
3540
.3627
.3714
3800
i
3885
3969
4053
4136
1219
.4301
. 4382
4462
.4542
4621
5
4099
4777
.4854
. 4930
. 5005
,5080
.5154
.5227
. 5299
5370
5441
551 I
5580
.5649
. 571 7
5784
5850
.5915
. 5980
. 6044
7
0107
.01(59
6231
.(')291
6351
.6111
6469
. 6527
6584
0640
8
(51.90
0751
6805
6858
.6911
6963
.7014
. 7064
.7114
7163
9
721 1
7259
. 7306 7352
. 7398
.7443
.7487
.7531
7574
7616
1
7658
7699
. 7739
7779
7818 | 7857
. 7895
. 7932
7969
8005
1 1
804 1
80 70
8110
8144
8178
8210
8243
. 8275
. 8306
8337
1 2
8367
8397
8426
8455
.8483
.8511
. 8538
. 8565
.8591
8617
i ,3
8043
8008
. 8692
.8717
.8741
8704
.8787
.8810
. 8832
8854
] 1
8875
8890
8917
8937
8957
. 8977
8996
.9015
9033
.9051
I 5
9069
9087
9104
.9121
9138
9154
9 1 70
9186
.9201
9217
1 (i
9232
924(5
9261
. 9275
. 9289
9302
9316
9329
. 934 1
9354
1 7
9360
.9379
.9391
. 9 102
. 94 1 4
9425
9436
.9417
9458
94681
1 8
94783
94884
91983
95080
.95175
95268
95359
. 95449
95537
95624
I 9
95709
95792
95873
95953
. 96032
96 1 09
96185
. 96259
.96331
96403
2.0
904 73
96541
96609
96675
96739
96803
. 96865
. 96926
96986
97045
2.1
97103
97159
97215
97269
97323
.97375
.97426
.97477
. 97526
. 97574
2 2
97022
97608
97714
97759
97803
97846
97888
97929
. 97970
98010
2 3
98049
98087
98124
,98161
98197
. 98233
98267
98301
98335
98307
2 4
98399
98431
98462
.98492
98522
.98551
. 98579
. 98607
98635
.98661
o r
98088
.98714
98739
98764
98788
98812
98835
. 98858
.98881
98903
2 6
98924
98945
98966
98987
. 99007
99026
.99015
99064
. 99083
99101
2.7
99118
99 1 36
99153
.99170
99180
99202
99218
. 99233
99248
. 99263
2 8
99278
. 99292
. 9J300
99320
99333
99346
. 99359
.99372
. 99384
. 99396
2.9
.99408
99420
99431
99443
.99454
. 99464
99475
99485
.99195
.99505
For great/or accuracy, arid for values beyond the table, r = (e iz 1 ) -f- (e~ z -\ 1);
z = KfloK (1 + r) - lug (1 - r)}
* RepiinU'd from Table V.B. of Fishei's "Statistical Methods for Reseat ch Workers,"
Oliver & Royd, Edinburgh, by kind peimiHHion of the author and pubhsheiN.
APPENDIX
419
TABLE V. SIGNIFICANT VALUES OF r AND 72. *
Values for P = .05 in lightfacc type. Values for P .01 in boldface type
Degi ee,B of
freedom
Number of variables
2
3
999
1.000
4
999
1.000
5
.999
1.000
7
9
13
25
1 000
1.000
1
907
1.000
1 000
1.000
1 000
1.000
1 000
1.000
1 000
1.000
2
950
.990
975
.995
983
.997
.987
.998
990
.998
992
.998
994
.999
990
.999
998
1.000
3
.878
.959
930
.976
950
.983
901
.987
.908
.990
973
.991
979
.993
980
.995
993
.998
4
.811
.917
881
.949
912
.962
.930
.970
.942
.975
950
.979
901
.984
973
.989
980
.994
5
. 754
.874
830
.917
874
.937
898
.949
.914
.957
.925
.963
.941
.971
958
.980
978
.989
6
707
.834
795
.886
839
.911
807
.927
.880
.938
.900
.946
920
.967
943
.969
909
.983
7
000
.798
758
.855
807
.885
838
.904
800
.918
876
.928
900
.942
927
.958
960
.977
8
032
.765
720
.827
777
.860
811
.882
835
.898
854
.909
880
.926
912
.946
950
.970
9
602
.735
(597
.800
750
.836
.780
.861
812
.878
832
.891
.801
.911
.897
.934
941
.963
10
570
.708
071
.776
720
.814
703
.840
.790
.859
812
.874
843
.895
882
.922
932
.955
11
553
.684
048
.753
703
.793
741
.821
.770
.841
792
.857
820
.880
80S
.910
922
.948
12
532
.661
<>27
.732
083
.773
722
.802
751
.824
774
.841
.809
.866
. 854
.898
913
.940
13
511
.641
008
.712
.004
.755
703
.785
.733
.807
.757
.825
.794
.852
.840
.886
904
.932
14
197
.623
590
.694
. 040
.737
. 080
.768
.717
.792
741
.810
779
.838
.828
.875
. 895
.924
15
482
.606
574
.677
. 030
.721
.070
.752
.701
.776
720
.796
705
.825
815
.864
880
.917
10
.408
.590
559
.662
.015
.706
055
.738
. 080
.762
712
.782
751
.813
.803
.853
.878
.909
17
.450
.576
545
.647
.001
.691
.041
.724
.073
.749
.098
.769
738
.800
792
.842
869
.902
18
.444
.561
. 532
.633
. 587
.678
.028
.710
. 000
.736
. 080
.756
.720
.789
781
.832
801
.894
19
. 433
.549
.520
.620
.575
.665
.015
.698
.047
.723
.074
.744
.714
.778
.770
.822
853
.887
20
. 423
.537
509
.608
. 503
.652
.004
.685
.036
.712
.002
.733
. 703
.767
.700
.812
. 845
.880
21
.413
.526
198
.596
. 552
.641
.592
.674
.024
t 700
.051
.722
.093
.756
.750
.803
.837
.873
22
404
.515
188
.585
542
.630
582
.663
.014
.690
040
.712
.082
.746
.740
.794
.830
.866
23
.390
.505
.479
.574
532
.619
.572
.652
004
.679
. 080
.701
073
.736
.731
.785
. 823
.859
24
. 388
.496
.470
.565
523
.609
. 502
.642
. 594
.669
.021
.692
.003
.727
.722
.776
.815
.852
* Reprinted by kind permission of Dr. George W. Snedecor from
Machine Calculation" (1931).
' Correlation and
420
METHODS OF PLANT BREEDING
TABLE V. -SicNiFirANT VALUES OF r AND R.* (Continued^
Values for P = .05 in lightfaoc type. Values for P = .01 in boldface type
Degrees of
freedom
_.
Number of variables
3
4
5
6
7
9
13
25
. __ ^
~381
.487
46 2~
.555
' "ini
.600
553"
.633
585
.660
.612
.682
654
.718
- 714-
.768
". 80 8 "
.846
26
374
.478
.454
.546
506
.590
545
.624
576
.651
.603
.673
645
.709
706
.760
.802
.839
27
367
.470
446
.538
,498
.682
. 536
.615
. 568
.642
594
.664
637
.701
698
.752
.795
.833
28
361
.463
439
.530
. 490
.573
529
.606
560
.634
586
.656
.629
.692
.690
.744
.788
.827
29
.355
.456
432
.522
.482
.665
521
.598
552
.625
.579
.648
.621
.685
.682
.737
.782
.821
30
349
.449
.426
.514
.476
.558
.514
.591
. 545
.618
571
.640
.614
.677
675
.729
776
.815
35
.325
.418
397
.481
445
.523
482
.556
512
.582
538
.605
580
.642
642
.696
746
.786
40
301
.393
373
.454
419
.494
455
.526
484
.552
. 509
.576
.551
.612
613
.667
720
.761
45
288
.372
353
.430
397
.470
. 432
.501
. 460
.527
.485
.549
.526
.586
. 587
.640
. 696
.737
50
273
.354
336
.410
. 379
.449
412
.479
440
.504
464
.526
.504
.562
. 565
.617
.674
.716
60
250
.325
308
.377
.348
.414
380
.442
.406
.466
.429
.488
.467
.523
.526
.577
.636
.677
70
232
.302
286
.351
324
.386
354
.413
379
.436
401
.456
438
.491
.495
.544
004
.644
80
.217
.283
.269
.330
.304
.362
. 332
.389
. 356
.411
.377
.431
.413
.464
.469
.516
576
.615
90
205
.267
254
.312
288
.343
.315
.368
338
.390
358
.409
392
.441
446
.492
552
.690
100
195
.254
241
.297
.274
.327
.300
.351
.322
.372
.341
.390
.374
.421
426
.470
530
.668
125
.174
.228
.216
.266
246
.294
.269
.316
. 290
.335
.307
.352
.338
.381
387
.428
485
.521
150
.159
.208
.198
.244
.225
.270
.247
.290
.266
.308
.282
.324
.310
.351
.356
.395
.450
.484
200
.138
.181
.172
.212
.196
.234
.215
.253
.231
.269
246
.283
.271
.307
312
.347
398
.430
300
.113
.148
.141
.174
J60
.192
.176
.208
.190
.221
.202
.233
.223
.253
.258
.287
.332
.359
400
.098
.128
.122
.151
. 1 39
.167
. 153
.180
.165
.192
.176
.202
.194
.220
.225
.250
291
.315
500
.088
.116
. 109
.135
.124
.150
.137
.162
.148
.172
.157
.182
.174
.198
.202
.225
. 262
.284
1000
.062
.081
.077
.096
.088
.106
.097
.115
.105
.122
.112
.129
.124
.141
.144
.160
.188
.204
For total and partial coirelation coefficients, use column for 2 variables in table for 5 and 1
per eerit points. Degrees of freedom = (number of observations) (number of variables).
For multiple-correlation coefficients, use the column corresponding to the number of
variables. Degrees of freedom = (number of observations) (number of variables).
* Reprinted by kind permission of Dr. George W. Snedecor from "Correlation and
Machine Calculation" (1931).
APPENDIX
421
TABLE VI. TRANSFORMATION OF PERCENTAGE TO DEGREES
Percentage (p) = sin 2 0*
Per cent
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.0
1 8
2 6
3 1
3 6
4 1
4 4
4.8
5 1
5 4
1
5 7
6
6 3
6.5
6 8
7.0
7 3
7 5
7 7
7 9
2
8 1
8.3
8 5
8 7
8 9
9 1
9 3
9 5
9.6
9 8
3
10.0
10.1
10.3
10.5
10.6
10.8
10.9
11.1
11.2
11 4
4
11.5
11.7
11.8
12.0
12.]
12.2
12 4
12.5
12.7
12 8
5
12 9
13 1
13 2
13.3
13.4
13.6
13.7
13.8
13 9
14 1
6
14 2
14.3
14.4
14 5
14 7
14.8
14 9
15.0
15 1
15 2
7
15 3
15 5
15 6
15 7
15 8
15 9
16
16 1
16 2
16 3
8
16 4
16 5
16 6
16 7
16 8
17
17.1
17.2
17 3
17 4
9
17.5
17.6
17.7
17 8
17 9
18
18
18.1
18 2
18 3
10
18.4
18 5
18.6
18.7
18.8
18.9
19.0
19.1
19.2
19 3
11
19 4
19 5
19.6
19.6
19.7
19.8
19.9
20
20.1
20.2
12
20 3
20 4
20.4
20.5
20.6
20.7
20 8
20.9
21.0
21.0
13
21 1
21 2
21.3
21 4
21.5
21.6
21.6
21.7
21.8
21.9
14
22
22 I
22 1
22.2
22.3
22 4
22 . 5
22 5
22.6
22 7
15
22 8
22 9
22 9
23.0
23.1
23.2
23.3
23 3
23.4
23.5
16
23 6
23 7
23 7
23.8
23.9
24.0
24
24.1
24.2
24 3
17
24 4
24 4
24.5
24 6
24.7
24.7
24.8
24.9
25
25
18
25 1
25.2
25 3
25.3
25.4
25 . 5
25 5
25 6
25 7
25 8
19
25 8
25 9
26
26 1
26.1
26.2
26 3
26.3
26 4
26 5
20
26 6
26 6
26 7
26.8
26.9
26.9
27.0
27.1
27.1
27 2
21
27 3
27 3
27.4
27 5
27.6
27 6
27.7
27 8
27.8
27 9
22
28
28
28 1
28.2
28.2
28 3
28 4
28.5
28 5
28 6
23
28.7
28.7
28 8
28 9
28.9
29.0
29 1
29 1
29 2
29 3
24
29 3
29.4
29.5
29 5
29.6
29.7
29.7
29.8
29.9
29.9
25
30
30.1
30.1
30.2
30.3
30.3
30.4
30.5
30.5
30.6
26
30.7
30 7
30.8
30.9
30.9
31.0
31.0
31.1
31.2
31 2
27
31 3
31.4
31.4
31.5
31.6
31.6
31.7
31.8
31.8
31 9
28
31 9
32.0
32 1
32 1
32.2
32 3
32.3
32 4
32.5
32 5
29
32.6
32 6
32.7
32.8
32.8
32.9
33.0
33.0
33.1
33 1
30
33 2
33.3
33.3
33.4
33.5
33.5
33.6
33.6
33.7
33 8
31
33 8
33.9
34.0
34.0
34.1
34.1
34 2
34.3
34.3
34 4
32
34 4
34.5
34 6
34.6
34.7
34.8
34.8
34.9
34.9
35
33
35.1
35.1
35.2
35 2
35.3
35.4
35.4
35.5
35.5
35 6
34
35 7
35 7
35.8
35 8
35.9
36.0
36
36.1
36.2
36 2
35
36.3
36.3
36 4
36.5
36.5
36 6
36 6
36.7
36.8
36 8
* Published by kind permission of Dr. C. I. Bliss (1937).
422 METHODS OF PLANT BREEDING
TABLE VI. TRANSFORMATION OF PERCENTAGE TO DEGREES. (Continued)
Per cent
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
36
36 9
36 9
37
37
37 1
37 2
37 2
37 3
37 3
37 4
37
37 5
37 5
37.6
37.6
37.7
37.8
37.8
37.9
37.9
38.0
38
38 1
38 1
38 2
38 2
38 3
38 4
38.4
38.5
38.5
38.6
39
38.6
38 7
38 8
38.8
38 9
38 9
39.0
39.1
39 1
39 2
40
39.2
39 3
39.3
39 4
39.5
39.5
39.6
39 6
39.7
39.8
41
39 8
39 9
39 9
40.0
40.0
40 1
40.2
40.2
40 3
40.3
42
40 4
40 5
40 5
40 6
40 6
40.7
40.7
40.8
40 9
40 9
43
41
41
41 1
41 1
41.2
41 3
41 3
41 4
41 4
41 5
44
41 6
41 6
41 7
41 7
41.8
41.8
41 9
42
42.0
42 1
45
42.1
42.2
42 2
42.3
42 4
42 4
42 5
42 5
42 6
42 6
46
42 7
42 8
42 8
42 9
42.9
43
43
43.1
43 2
43 2
47
43 3
43 3
43 4
43 5
43 5
43 6
43
43 7
43 7
43 8
48
43 9
43 9
44
44
44 1
44 1
44 2
44 3
44 3
44 4
49
44 4
44 5
44 5
44 6
44 7
44.7
44.8
44 8
44 9
44 9
50
45
45.1
45.1
45.2
45.2
45 3
45 3
45 4
45 5
45 5
51
45.6
45 6
45.7
45.7
45 8
45 9
45 9
46.0
46
46 1
52
46 1
46 2
46 3
46 3
46.4
46.4
46.5
46 . 5
46 6
46 7
53
46 7
46 8
46 8
46 9
47
47
47.1
47 1
47 2
47 2
54
47.3
47 4
47.4
47.5
47 5
47 6
47.6
47 7
47 8
47 8
55
47.9
47 9
48
48.0
48.1
48.2
48.2
48.3
48.3
48 4
56
48 4
48 5
48 6
48.6
48.7
48 7
48 8
48.9
48 9
49
57
49
49 1
49 1
49 2
49.3
49.3
49 4
49.4
49.5
49 5
58
49 6
49 7
49 7
49.8
49 8
49 9
50.0
50.0
50.1
50 1
59
50 2
50 2
50 3
50 4
50.4
50.5
50.5
50 6
50.7
50.7
60
50.8
50.8
50 9
50.9
51.0
51.1
51.1
51.2
51.2
51 3
61
51 4
51 4
51.5
51 5
51.6
51.6
51.7
51.8
51.8
51.9
62
51 9
52
52 1
52.1
52.2
52 2
52.3
52.4
52.4
52.5
63
52 5
52 6
52 7
52 7
52.8
52.8
52.9
53.0
53.0
53 1
64
53.1
53.2
53.2
53 3
53.4
53.4
53.5
53.5
53 6
53 7
65
53.7
53 8
53.8
53.9
54.0
54.0
54.1
54.2
54.2
54.3
66
54.3
54 4
54 5
54.5
54 6
54 6
54.7
54.8
54.8
54.9
67
54 9
55.0
55 ]
55 1
55 2
55 2
55.3
55.4
55 4
55.5
68
55 6
55.6
55 7
55.7
55 8
55 9
55 9
56
56.0
56 1
69
56.2
56 2
56.3
56.4
56.4
56.5
56.5
56.6
56.7
56.7
70
56.8
56.9
56.9
57.0
57.0
57.1
57.2
57.2
57.3
57.4
APPENDIX 423
TABLE VI. TRANSFORMATION OF PERCENTAGE TO DEGREES. (Continued}
Per cent
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
71
57.4
57.5
57.5
57.6
57.7
57.7
57.8
57.9
57 9
58
72
58.1
58.1
58.2
58.2
58 3
58.4
58.4
58.5
58.6
58 6
73
58 7
58.8
58 8
58 9
59
59. (
59 1
59 . 1
59.2
59 3
74
59 3
59 4
59 . 5
59 5
59 . (
59.7
59.7
59.8
59 9
59 9
75
60
60 1
60.1
60 2
60.3
60 3
60.4
60 . 5
60 5
60 6
76
60 7
60.7
60.8
60.9
60 9
61.0
61.1
61.1
61 2
61 3
77
61 3
61 4
61 5
61 5
61
61.7
61 8
61 8
61.9
62
78
62
62 . 1
62.2
62 2
62 . 3
62 4
62 4
62.5
62.6
62.7
79
80
62 7
63 . 4
62 . 8
63 5
62.9
63.6
62.9
63.7
63 .
63 . 7
63.1
63.8
63 1
63 9
63 . 2
63 9
63 3
64
63 4
64 1
81
64.2
64.2
64.3
64.4
64.5
64.5
64.6
64.7
64.7
64 8
82
64 9
65
65
65 1
65 . 2
65 3
65 . 3
0.5 4
65 5
65 6
83
05
65 . 7
05 8
65 . 9
66.0
60
66 1
66.2
66 3
66 3
84
66 4
66 . 5
66 6
GO 7
66.7
66.8
66.9
67.0
67.1
67 1
85
07 2
67.3
67 4
67 5
67.5
67.6
67 7
67 8
67.9
67.9
86
87
68
08 9
08 1
09
08 2
09
68 3
69 1
08 4
09 2
68.4
69 3
68.5
09 4
68
69 5
68 7
69.6
68.8
69 6
88
09 7
69.8
09 9
70
70 1
70.2
70 3
70.4
70.4
70 5
89
70
70 7
70 8
70 9
7J .0
71 I
71.2
71.3
71 4
71 5
90
71 6
71 7
71.8
71 9
72.0
72
72.1
72.2
72.3
72 4
91
72 5
72
72 7
72 8
72 . 9
73.0
73 2
73 3
73 4
73 5
92
73 6
73 7
73 8
73 9
74.0
74 1
74 2
74 3
74 4
74 5
93
74.7
74 8
74 9
75
75 . 1
75.2
75 3
75 5
75 6
75.7
94
75 8
75 9
76 1
76 2
76 . 3
76.4
76 .
76 7
76.8
76 9
95
77.1
77 2
77.3
77.5
77.6
77.8
77.9
78.0
78 2
78 3
96
78.5
78
78. S
78 9
79.1
79 2
79.4
79.5
79.7
79 9
97
80.0
80.2
80.4
80 5
80 7
80.9
81.1
81.3
81.5
81.7
98
81.9
82.1
82 3
82 5
82 7
83
83 2
83 5
83 7
84
99
84.3
84.6
84.9
85.2
85 6
85 9
86 4
86 9
87 4
88 2
100
90
INDEX
Aamodt, 134-136, 138
Aase, 20
Ahlgren, 185
Alexander, 188
Alfalfa, effects of self-fertilization,
243-247
inflorescence, 68-69
selfing and crossing, 69
Allard, 85
Anderson, 203
Arny, 170, 171, 297, 298
Ashhy, 54
Atkins, 180
Atwood, 69, 249
Ausermis, 139, 331, 332
B
Barker, 173
Barley, artificial epiphytotics, fusar-
ial head blight, 120
smuts, 120
stem rust, 118
breeding, 88-91
chromosome numbers, 155
cross-pollination, 42
crossing methods, 63
effects of selling, 99-100
inheritance, internode length, 155
quantitative characters, 163
reaction to Ilelminthosporium
nativum, 159
reaction to mildew, 162
reaction to stem rust, 160
in species crosses, 152-153
linkage groups, 155-156
species, 152-153
Bartlett, 341
Bateson, 11
Baur, 114
Beadle, 224
Beans, mosaic resistance, 115
Beddows, 249
Biffin, 132, 133
Bindloss, 56
Blakeslee, 34-36
Blanchard, 241
Bliss, 375, 421
Blodgett, 41
Bolley, 114, 172
Borgeson, 268
Boyack, 132, 133
Bressman, 218
Brewhaker, 194, 247, 301
Brieger, 251
Bnerley, 35
Bnggs, 104, 139, 140, 162
Brink, 255
Broadfoot, 171
Brookms, 89, 117, 127, 162
Brtmson, 189, 197, 236, 239
Bryan, 209, 237, 238, 304
Burlison, 239
Burnham, 110, 173, 224, 235
Bushnell, 250
Cabbage, wilt resistance, 115
yellows, 126
Campbell, 182
Cantaloupe, resistance to mildew,
104
Cartledge, 1 10, 235
Cartter, 85
Chang, 178, 179
Christidis, 303
Chromosome numbers, in cereals, 1?
425
426
METHODS OF PLANT BREEDING
Chromosome numbers, in fiber
plants, 13
in* forage grasses, 12-13
m fruits, 14
in legumes, 13
in oats, 141
in oil plants, 13
in stimulants, 13
in sugar plants, 13
in vegetables, 14
Churchward, 139
Clark, Andrew 376
Clark, E. R., 181
Clark, J. A., 83, 131, 132, 139
Clarke, 247
Clayton, 85, 182
Cochran, 292, 352, 367, 375
Coffman, 143, 150
Collins, G. N., 52, 241
Collins, J. L., 40
Combining ability, in corn, 236-239
in small grains, 98
Cooper, D. C., 255
Cooper, II. P., 134
Corn, artificial epiphytotics, smuts,
121, 122
breeding improved inbred linos,
by baekcross method, 210
by convergent improvement,
52-53, 210-214
by pedigree method, 207
breeding methods, 6-8, 187-214
combining ability, 236-239
controlled pollination methods,
193-214
drought resistance, 181
effects of selfing, 48-50
inheritance, of chlorophyll varia-
tions, 221
of endosperm characters, 218
of glossy seedlings, 223
of plant colors, 222
linkage map, 225
linkage studies, 224-231, 233-236
origin and classification, 215-218
pericarp tenderness, 108, 240
protein content, 239
quantitative inheritance, 231
Corn, resistance of, to bacterial wilt,
241
to cold, 239
to drought, 239
to insects, 241
to rust, 241
to smut, 109, 234
seed increase of inbreds and F\
crosses, 268-269
selection without controlled pol-
lination, 187-191
selfing and crossing technics, 61
Cotton, effects of self-pollination, 47
selfing and crossing, 64
Cox, 352, 367
Craig, 81, 145, 148
Craigie, 136
Crane, 251
Cummmgs, 48, 250
Curtis, 250
(Hitler, 176
Cytogenetics, nllopolyploids, 16
autopoiyploids, 16
D
Dahms, 241
Darlington, 130
Darwin, 242
Davis, 236
Derm an, 34
Dickson, 135, 148
Dillman, 85, 165, 167, 169
Disease resistance, in barley, 90
in beans, 115
in cabbage, 115
types of, 126
in cantaloupe's, powdery mildew,
104
in corn, smut, 109
in flax, wilt-, 114
in grapes, 1 1 4
importance of, 113-116
methods of breeding for, 122 124
nursery methods, 82
in oats, crown rust, 27, 149
smuts, 29, 150
stem rust, 28, 148
INDEX
427
Disease resistance, production of
epiphytotics, 118-122
reaction to bacterial wilt of corn,
241
reaction to Hehninthosporium sati-
vum, 159
reaction to insect attacks, 241
reaction to rust of corn, 240
reaction to smut of corn, 234
in snapdragons, rust resistance,
106
in tomatoes, 115
in vegetables, 115
in watermelons, 115
in wheat, bunt, 104
Hessian fly, 127
leai rust, 107
mature-plant resistance, 4, 117,
127
physiological resistance, 4, 117,
126
stem rust, 107
Doxtator, 202
E
East, 45, 49-51, 53, 77, 193, 218,
221, 232, 239, 242, 251, 254
Eckhardt, 209, 237, 352, 307
Edgerlon, 1J5
Ellerton, 131
Emerson, 219, 222, 224, 231, 232
Emsweller, 35, 60, 106, 251
Engledow, 130, 133, 152
F
Fairer, 87, 139
Fernow, 41
Field plot methods, Latin squares,
312-313
lattice, simple, 351-365
triple, 365 -375
multiple experiments, 339-350
randomized blocks, 307-312
replication, 304
size and shape of plots, 302
split-plot designs, 315-320
yield trials in Minnesota, 304-306
Field plot teehnic, competition, 297
crop rotation, 289-291
soil heterogeneity, 291
Fisher, 287, 295, 307, 308, 312, 322,
331, 375, 411, 417, 418
Flax, artificial epipliytotic, rust, 120
wilt, 120
chromosome numbers, 165
crossing methods, 64
dehiscence of bolls, 168
inflorescence, 166
inheritance, of flower and seed
color, 167
of quality of oil, 170
of resistance to rust, 174
of resistance to wilt, 171
smooth vs. cihate septa, 169
of weight of seed and oil con-
tent, 169
of wilt resistance, 114, 171
Flor, 174
Fraps, 219
Eraser, 143, 145, 224
Freeman, 133
G
Gaines, 2, 20, 134, 139, 147
Garber, 39, 60, 147, 157, 158, 218,
236, 239, 293, 296
Garl, 70
Garner, 85
Garrison, 189, 190
Genetics, application of, to plant
breeding, 25-34
of barley, 152-164
of bunt resistance, 140
of flax, 165-174
genomes in wheat, 20-21
of maize, 215-241
of oats, 141-151
of plant pathogens, 125-126
of self-incompatibility, 251-257
of wheat, 12&-140
Oilman, 115
Glossary, 381-389
Goulden, 81, 136, 139, 351, 363
Grapes, resistance of, to vine louse
and vine mildew, 114
428
METHODS OF PLANT BREEDING
Grasses, iffects of selfing, 249
selfing, methods, 71
Griffee, 89, 160, 161
II
Haber, 239, 250
Hall, A. D., 281
Hall, D. M., 180
Hamilton, 247
Harlan, 84, 88, 97, 98, 101, 152, 153,
155, 157
Harrington, 98, 99, 136, 179, 180
Harris, 292, 293, 296, 320
Harvey, 240
Hauge, 219
Hayes, 5, 23, 39, 49-51, 54, 60, 88,
89, 108, 135, 138, 143, 144, 145,
150-153, 157, 158, 188, 193, 199,
207, 208, 218, 221, 235, 236,
239, 240, 268 7 293, 297, 298,
308, 323, 339
Hays, 76
Henkemeyer, 133
Henry, 174
Heribert-Nilsson, 247
Heterosis, explanation of, 50-56
Heyne, 181, 239
Hines, 89, 160
Hitchcock, 216
Hogg, 185
Holbert, 180, 239
Hollowell, 69
Hoover, 234, 296
Hopkins, 187
Hor, 152
Howards, 131, 132
Humphrey, 47, 65
Humphries, 176
Hunter, H., 57
Hunter, J. W., 181
Huskins, 24
Hutchins, 250
Immer, 99, 155, 234, 292, 299, 301,
308, 322, 339
Ivanoff, 241
Jagger, 250
Jenkm, 71, 73, 247, 249
Jenkins, E. W., 48, 250
Jenkins, M. T., 61, 195, 197, 201,
202, 206, 236, 237, 239
Jodon, 65
Johaimseii, 76
Johnson, I. J., 54, 108, 169, 199, 202,
207, 208, 236, 239, 240
Johnson, T., 135, 138
Jones, D. F., 42, 45, 46, 49, 51, 52,
193, 206, 218, 236, 239, 242
Jones, H. A., 66, 104, 106
Jones, J. W., 85
Jones, L. II., 115
Jorgenson, 194
K
Kearney, 47
Kechle, 50
Kempton, 240, 241
Kerns, 40
Kezer, 132, 133
Kicsselbach, 55, 187, 195, 298-300,
302
Kihara, 20, 141
Kirk, 69, 71, 244, 249
Knowlcs, 179, 180
Kostoff, 20
Kouznetsov, 134
Kurtzweil, 138
L
Larmour, 182
Larson, 179
Latin squares, 312-313
Laudc, 181
Lawrence, 251
Leake, 57
Leith, 89
Leonard, 153, 376
Levy, 259
Lilienfeld, 20
Lindstrom, 55, 233
INDEX
429
Link, 185
Love, H. H., 81, 145, 147, 148
Love, R. M., 23
Luckwill, 55
M
Macaulcy, 259
McFadden, 6, 22, 138
Mclllvaine, 296
Macindoe, 137, 138
McRostie, 147
Mac Vicar, 185
Mains, 240
Mangelsdorf, 215, 216, 219, 221
Marston, 241
Martini, 84, 98
Mayer, 194
Mendel, 11, 86
Mercer, 281
Meyers, 241
Missing plots, estimating yield of,
314
Montgomery, 187
Moore, 23, 143, 144, 323
Morse, 85
Muller, 2
Murphy, H. C., 150
Murphy, R. P., 213
Mutations, induced in wheat, oats,
and barley, 21
nondefective, 78
Myers, 22, 169, 174, 241
N
Neal, 195
Neatby, 136, 139
Nebel, 36
Newman, 75
Newton, 135, 138
Nilsson, H. N., 75
Nilsson-Ehle, 133, 134, 145, 148
Nilsson-Leissner, 134, 194
Nishiyama, 141, 143
Noll, 81
Nowosad, 185
O
Oats, artificial epiphytotics, crown
rust, 118
smuts, 120
stem rust, 118
breeding of, 25-34, 92
crossing methods, 63
fatuoids, 24
inflorescence, 146
inheritance, of awn development,
144
of byzantina-sativa characters,
23, 30-33
of color of grain, 145
of crown rust reaction, 27, 149
of hulled vs. hull-less, 147
of pubescence, 148
of quantitative characters, 151
of smut reaction, 29, 150
of spreading vs. side panicle, 147
of stem rust reaction, 28, 148
loose smut, 124
quantitative characters, 33, 151
species crosses, 141-144
stem rust, 28, 117, 123
Onion, selfing and crossing, 66
Painter, 127
Pan, 139
Parker, J. H., 138, 143
Parker, W. H., 134
Pearl, 3
Pearson, Karl, 283
Pearson, O. H., 258
Pellew, 50
Pelskcnkc, 176
Pcrcival, 132
Peterson, 138, 247
Philp, 24, 143, 147, 148
Physiologic races, Hessian fly, 122
loose smut of oats, 124
rust in flax, 174
stem rust, of oats, 123
of wheat, 4-6, 123, 135
wilt of flax, 171
430
METHODS OF PLANT BREEDING
Pieters, 69
Plant breeding, applications of gene-
tics, 25-34
colchicine in, 34-38
hybridization methods in self-
pollinated plants, 86-100
outline of procedure, 91-98
important phases, 1
methods, asexual groups, 39-41
backcross, 97, 101-112
bulk, 96
classification, 56
convergent improvement, 52-53
corn breeding, 6-8, 187-214
cross-pollinated plants, 257262
disease and insect resistance,
113-128
multiple crosses, 97
pedigree, 95
potato improvement, 8-10
pure line, 74-85
outline of procedure, 79-83
in Scandinavia, 75
sexual group, 41-50
technics in selfing and crossing,
60-73
value of, 2
Vilmorin's isolation principle,
75
polyploids in relation to, 15-25
selection methods, for cold re-
sistance, 177
for oouraarin content in sweet
clover, 182
for dormancy in wheat, 179
for drought resistance in corn,
181
for HCN in sudan grass, 185
for lodging resistance in small
grains and corn, 180
for quality in wheat, 175
for shattering resistance in
wheat, 178
for wheat-meal fermentation
time test, 176
of wheat, rust resistance, 3-6
Polyploids, autopolyploid inherit-
ance, 18-19
Polyploids, origin of, 16
in relation to plant breeding,
15-25
Pomeroy, 40
Poole, 250
Pope, 101
Porter, 250
Potatoes, breeding methods, 8-10
crossing methods, 66
Powers, 22, 23, 43, 89, 135, 160, 163,
164, 308, 339
Prell, 251
Pumpkin, selfing and crossing, 66
Punnett, 11
Q
Quinby, 65
Quiscnbcrry, 131, 132, 143, 177, 236
R
Raleigh, 292
Randomized blocks, 307-312
multiple experiments, 339-350
Rasmussen, 161
Red clover, effects of sclfing, 249
selfing and crossing methods, 67
Reed, 124, 125, 148, 150
Reeves, 215, 216
Reid, 89, 161
Rice, crossing methods, 65
Richey, 52, 53, 102, 189-191, 194,
195, 197, 210
Rider, 332
Hi GUI an, 115
Riker, 241
Riley, 253, 254
Roberts, 185
Robertson, 152, 153, 155
Rosa, 250
Ruttle, 36
Rye, effects of selfing, 247
selfing and crossing methods, 64
Saboe, 235
Salmon, 180, 376
INDEX
431
Sansome, 24
Saunders, 176
Scofield, 296
Scott, G. W., 250
Scott, L. B., 39
Seed certification, 263-279
Canadian Seed Growers' Associa-
tion, 265, 269
increase by Minnesota method,
266
International Crop Improvement
Association, 270
Minnesota method, 272-277
potatoes, 277
Maine method, 278- 279
recommended varieties, 264
selecting the variety, 263
Self-fertilization, effects of, in al-
falfa, 244-247
in corn, 48, 242
in cotton, 47
in cross- pollinated plants, 48
50
formulas, 4546
in often cross-pollinated plants,
16-47
in rye, 247
in sunflowers, 247
in timothy, 247
Shamel, 39, 40
Shaw, 168
Shull, 50, 192, 193
Singleton, 206
Smith, D. C., 117, 123, 127, 148, 150
Smith, H. H., 36
Smith, L. II., 189
Snapdragons, rust resistance, 106
tetraploid, 35
Snetlecor, 310, 335, 375, 412, 419
Snelling, 241
Sorghum, artificial epiphytotics,
smut, 121-122
crossing methods, 65
Spillman, 134
Sprague, 52, 53, 55, 210, 238
Squash, selfing and crossing meth-
ods, 66
Stadler, 21
Stakman, 23, 125, 126, 135, 138, 143,
144, 323
Stanford, 162
Stan ton, 84, 92, 141, 150
Statistics, applied to data in per-
centages, 375-379
Chi square tests, 320-324
comparison of varieties through
check, 350-351
correlation, multiple, 338
partial, 333-338
simple, 284, 325-329
means and differences of,
331-333
definition of constants, 280-281
mean, mode, standard error, vari-
ance, coefficient of variation,
281-284
regression, 329-331
t test, 285-288
tables of P\ 412-416
tables of p = sin 2 0, 421-423
tables of r and R, 419-420
tables of r to z, 418
tables of /, 411
tables of x 2 , 417
Stephens, 65
Stevenson, F. J., 42
Stevenson, T M., 69, 182, 185
Sturtevant, 217
Summerby, 297
Suneson, 64
Sunflowers, effects of selfing, 247
Surface, 143, 145, 148
Sweet clover, coumarin content, 182
inducing biennial habit, 182
selfing and crossing methods, 69
T
Tammes, 165, 167
Tedin, 307
Timothy, selfing effects, 247
Tmgcy, 92
Tippett, 308
Tisdale, 171
Tomatoes, cross-pollination, 42
wilt resistance, 115
432
METHODS OF PLANT BREEDING
Torrie, 143, 144, 147, 150
Trost, 219
Tuller, 65
Tysdal, 70, 244, 246, 299
Valle, 247
Vavilov, 34, 88, 134, 215
Vegetables, disease resistance, 115
Vilmorin, 75
Vogel, 178
W
Wade, 85
Walker, 126
Wallace, 218, 335
Wang, 56
Warmke, 35
Watermelons, wilt resistance, 115
Watkins, 129, 131
Weibel, 177
Weihing, 302
Wellhausen, 241
Welsh, 136, 139, 148, 149
Wclton, 189
Wexelsen, 159
Whaley, 56
Wheat, artificial epiphytotics, bunt,
119
fusarial head blight, 120
Hessian fly, 122
leaf rust, 118
stem rust, 118
bunt resistance, 104
crossing methods, 63
Wheat, cross-pollination, 43
genomes, 20-21, 129
germinal instability, 22-23
Hope and H44, 6, 23
inflorescence, 62
inheritance, of awnedness, 130
of bunt resistance, 139
of chaff characters, 132
of glume shape, 129
of quantitative characters, 140
of seed characters, 133
of spike density, 134
of spring vs. winter habit, 134
of stem rust reaction, 135
kernel color, 17
leaf rust, 107
resistance to Hessian fly, 127
spcltoids, 24
Thatcher, 5-6
Whitaker, 250
White, 185
Wiebe, 155, 352, 356, 365
Wiener, 269
Williams, C. G., 187, 189
Williams, R. P., 67, 249
Wilson, 181
Wmge, 19, 24
Woodward, 92
Worzella, 176
Wright, 45
Wu, 208, 236
Yasuda, 255
Yates, 308, 312, 314, 351, 352, 362,
375
TVNews
OpenLibrary